Association of weight-adjusted waist index (WWI) and a body shape index (ABSI) with serum neurofilament light chain levels in a national study of U.S. adults Authors: Zixuan Yan, Qingxin Gu, Hong Yin, Mingliang Yi, Xiao Wang, Ruirui Sun, Fanrong Liang, Dingjun Cai, Wenchuan Qi Published: 2024 ABSTRACT ObjectiveThis study explored how the weight-adjusted waist index (WWI) and a body shape index (ABSI) are related to serum neurofilament light chain (sNfL) levels among U.S. adults. We aimed to evaluate sNfL, which plays key roles in neuronal injury in neurological diseases, given its understudied connection to obesity.MethodsWe used cross-sectional data from the 2013–2014 National Health and Nutrition Examination Survey (NHANES) of people with complete information on the weight-adjusted waist index (WWI), a body shape index (ABSI), and serum neurofilament light chain (sNfL). Multiple linear regression analysis allowed us to investigate the separate connections among the WWI, ABSI, and sNfL. Moreover, interaction testing and subgroup analysis were performed to improve the general validity of our results. To assess any nonlinear correlations, we also performed threshold effect analysis and smoothed curve fitting.ResultsWWI and ABSI were positively linked with sNfL (WWI: β = 0.05, 95% CI 0.01–0.09; ABSI: β = 1.65, 95% CI 3.53–13.72). There was no clear reliance on this association according to subgroup analysis and interaction tests. Smoothed curve fitting and saturation effects also demonstrated nonlinear associations between WWI and ABSI and sNfL, with inflection points of 10.38 and 0.38, respectively.ConclusionIn the adult American population, while the WWI and ABSI are linearly positively correlated with serum neurofilament light protein (sNfL), the effect size is greater for the ABSI. This correlation provides fresh evidence connecting obesity to neurological conditions, deepening our comprehension of the extensive health impacts associated with obesity.Level of Evidence Level I, experimental studies. Introduction The biomarker serum neurofilament light chain (sNfL) is a vital sign of neuronal injury for a range of neurological diseases [1]. Neurofilaments are intracellular cytoskeletal proteins comprising light, medium, and heavy chain types, categorized by their molecular weight. These proteins are released into the extracellular milieu due to neuronal injury, and they may subsequently infiltrate the cerebrospinal fluid and blood circulation. With advancements in research, sNfL has been frequently termed the “neurologist’s C-reactive protein” [2]. Moreover, the detection of sNfL can facilitate the early diagnosis of neurodegenerative diseases, and there is substantial evidence supporting its role in assessing disease activity and progression in multiple sclerosis, Alzheimer’s disease, and ischaemic stroke [3–5]. Therefore, sNfL holds significant potential for a wide range of clinical applications. Given that elevated all-cause mortality rates are linked to obesity [6], there has been increasing focus on the relationship between obesity and cognitive health. Body mass index (BMI) has historically been used to describe obesity, and although abdominal (visceral) obesity leads to harmful metabolic disorders, subcutaneous fat accumulation has a benign effect on cardiometabolic risk. Increasing evidence paradoxically links increased BMI with better outcomes in patients with established cardiovascular disease, a finding known as the 'obesity paradox' [7]. However, previous studies have revealed an independent association between BMI and long-term mortality after ischaemic stroke only in underweight patients, and the beneficial effects of overweight and obesity were not detected. This suggests the limitation of the paradoxical phenomenon of obesity [8]. Another finding suggests that high BMI in middle-aged individuals seems to be a risk factor for dementia, but high BMI in later life sometimes seems to be protective. This result may help explain the ‘obesity paradox’ [9]. Because of the lack of studies on obesity and neurological diseases, the relationship between obesity and sNfL deserves further exploration. With the introduction of new obesity indices, such as the weight-adjusted waist index (WWI) and a body shape index (ABSI), studies on the associations between obesity and various diseases have increased in number. The WWI, an anthropometric indicator that focuses on central obesity, can be used to evaluate waist circumference more precisely to assess health risks [10, 11]. Research indicates that the obesity paradox, which relates BMI to mortality, does not extend to the association between the WWI and mortality [11]. ABSI, which is formulated via mortality statistics, serves as an abdominal obesity metric [12]. The ABSI stands out as the sole abdominal obesity index unaffected by the obesity paradox [13]. However, few existing studies have examined the correlations between the WWI and ABSI in relation to sNfL. Therefore, we examined potential links between these parameters using NHANES data. Materials and methods Data source Under the direction of the Centers for Disease Control and Prevention (CDC), the National Center for Health Statistics (NCHS) conducts an ongoing transversal survey known as the National Health and Nutrition Examination Survey (NHANES). It offers a wealth of resources including operational manuals, estimation techniques, consent documentation, and analytical protocols. To obtain comprehensive information on the NHANES operational guidelines, estimation techniques, permission forms, and data analysis processes, visit the NHANES website at https://wwwn.cdc.gov/nchs/nhanes/. The NCHS Ethics Review Board authorized the ethical conduct of the NHANES, and each participant provided written consent to take part in the study. Participants Given the scarcity of sNfL data, our analytical focus was narrowed to the 2013–2014 iteration of the NHANES. From the original cohort of 10,175 participants in this survey, we first excluded those with missing serum neurofilament light protein (sNfL) data (n = 8104). We determined the weight-to-waist index (WWI) using weight and waist circumference measurements and excluded individuals lacking complete WWI data (n = 72). Participants lacking ABSI information were also excluded (n = 2). Participants missing covariate information were also excluded, although the sample size was small. The missing information included self-reported smoking status (n = 1) and congestive heart failure (n = 1). Ultimately, 1,995 participants were enrolled in the research. The process of participant selection is schematically represented in Fig. 1.Fig. 1Flowchart for participants Flowchart for participants Independent and dependent variables Our study included WWI and ABSI as exposure variables. The raw data for WWI and ABSI were not directly provided by the NHANES program. Rather, waist circumference (cm) was divided by weight (kg) squared to obtain the WWI [14]. The ABSI involves dividing WC by (BMI^(2/3) × height^(1/2)) [12]. Serum neurofilament light protein (sNfL) was the dependent variable. In alignment with prior research protocols [15], serum samples collected between 2013 and 2014 were subjected to neurofilament light protein (NfL) analysis using a Siemens Healthineers-developed immunoassay that boasts exceptional sensitivity. This analysis was efficiently conducted on the fully automated Atellica immunoassay platform. For comprehensive information and detailed methodological protocols, kindly refer to the following URL: https://wwwn.cdc.gov/Nchs/Nhanes/2013-2014/SSSNFL_H.htm. Covariates To account for these modifications, we incorporated numerous confounding variables into the model, which may affect the correlation between the two types of variables, based on prior research [16, 17]. In general, these covariates were obtained from demographic information, self-reported survey data, and laboratory test results. The demographic data included age, sex, race, and the poverty-to-income ratio (PIR). The self-report questionnaire data included information on smoking status, stroke and congestive heart failure. The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation was used to estimate the glomerular filtration rate (eGFR) [18]. The blood volume of males and females was calculated using Nadler's formula on weight and height. Statistical analysis Statistical evaluation in this survey was performed using R (http://www.r-project.org) and EmpowerStats (http://www.empowerstats.com), with a significance threshold of p < 0.05. In addition, sampling weights (WTSSNH2Y) were used to be population representative using NHANES data. Whereas categorical variables are shown as frequencies with their associated percentages, variables reflecting a normal distribution are presented as the means with accompanying standard deviations (SDs). Natural logarithm transformation (ln-transformation) enables sNfL levels to reach approximation normalcy [17]. To evaluate the linkage between the predictor variables and the outcome variable, multivariable linear regression analysis was performed with adjustments for covariates in the model. Given the continuous nature of the variables, the subjects were initially categorized into Group 1 (sNfL level < 2.49 pg/ml) and Group 2 (sNfL level > 2.49 pg/ml) on the median ln-transformed sNfL level (2.49 pg/ml) [19]. Various models were developed depending on adjusted covariates, including Model 1 (unmodified covariates), Model 2 (modified for age, sex, and race), and Model 3 adjusted for all the covariates listed in Table 1, except WWI and the ABSI. This allowed us to reveal the association between the predictor and the result variable. To confirm these findings, subgroup analyses were also performed. To confirm the validity of a nonlinear interaction of the two sets of variables, we implemented threshold effect analyses and smoothing curve fitting techniques. These approaches offered further insights into the nature of the correlation.Table 1Baselines characteristics of participantsCharacteristicGroup 1Group 2P-valueParticipants, n9911004Sex (%) < 0.01 Male45.9452.17 Female54.0647.83Age (years) < 0.01 < 6094.1063.63 > = 605.9036.37Race < 0.01 Mexican American12.776.51 Other Hispanic6.395.40 Non-Hispanic White58.1671.59 Non-Hispanic Black14.239.57 Other race—including multi-racial8.446.93PIR (%)2.82 ± 1.652.94 ± 1.660.0982Smoking status, n (%) < 0.01 Yes38.3448.23 No61.6651.77Stroke, n (%) < 0.01 Yes0.943.85 No99.0696.15Congestive heart failure, n (%) < 0.01 Yes0.963.29 No99.0496.71eGFR (ml/min/1.73m2)116.79 ± 12.21105.02 ± 12.35 < 0.01BV(L)5.21 ± 0.945.15 ± 0.900.2178WWI10.78 ± 0.7711.10 ± 0.79 < 0.01ABSI0.40 ± 0.040.41 ± 0.04 < 0.01*Mean ± SDs for continuous variables: P-value was calculated by weighted linear regression model*% for categorical variables: P-value was calculated by weighted Chi-square test*Group 1: serum neurofilament light chain (sNfL) < 2.49 pg/ml; Group 2: serum neurofilament light chain (sNfL) ≥ 2.49 pg/ml Table 1Baselines characteristics of participantsCharacteristicGroup 1Group 2P-valueParticipants, n9911004Sex (%) < 0.01 Male45.9452.17 Female54.0647.83Age (years) < 0.01 < 6094.1063.63 > = 605.9036.37Race < 0.01 Mexican American12.776.51 Other Hispanic6.395.40 Non-Hispanic White58.1671.59 Non-Hispanic Black14.239.57 Other race—including multi-racial8.446.93PIR (%)2.82 ± 1.652.94 ± 1.660.0982Smoking status, n (%) < 0.01 Yes38.3448.23 No61.6651.77Stroke, n (%) < 0.01 Yes0.943.85 No99.0696.15Congestive heart failure, n (%) < 0.01 Yes0.963.29 No99.0496.71eGFR (ml/min/1.73m2)116.79 ± 12.21105.02 ± 12.35 < 0.01BV(L)5.21 ± 0.945.15 ± 0.900.2178WWI10.78 ± 0.7711.10 ± 0.79 < 0.01ABSI0.40 ± 0.040.41 ± 0.04 < 0.01*Mean ± SDs for continuous variables: P-value was calculated by weighted linear regression model*% for categorical variables: P-value was calculated by weighted Chi-square test*Group 1: serum neurofilament light chain (sNfL) < 2.49 pg/ml; Group 2: serum neurofilament light chain (sNfL) ≥ 2.49 pg/ml Baselines characteristics of participants *Mean ± SDs for continuous variables: P-value was calculated by weighted linear regression model *% for categorical variables: P-value was calculated by weighted Chi-square test *Group 1: serum neurofilament light chain (sNfL) < 2.49 pg/ml; Group 2: serum neurofilament light chain (sNfL) ≥ 2.49 pg/ml Results Characteristics of participants A total of 1,995 subjects were included in the research. As shown in Fig. 2, the sNfL levels did not follow a normal distribution; instead, on the logarithmic scale, they displayed a log-normal distribution. In addition, we found that some subjects' sNfL values were between 50 and 100 pg/ml. Currently, existing studies have analysed the normal reference value of sNfL, and 53.8 pg/mL is the maximum cut-off value of sNfl in normal people [20]. Therefore, we inferred the subjects with concentrations ranging from 53.8 to 100 pg/mL had a greater risk of neuroaxonal damage. When grouped by the median ln-transformed sNfL level (2.49 pg/ml), participants with higher ln-transformed sNfL levels also presented higher WWI (11.10 ± 0.79) and ABSI (0.41 ± 0.04) (p < 0.001). Additionally, participants with higher ln-transformed sNfL levels were more likely to be male (52.17%) and older (51.92 ± 14.51 years). Table 1 shows the characteristics of the participants. Participants with higher levels of ln-transformed sNfL were more likely to have smoked, had a stroke and congestive heart failure and had a lower eGFR (p < 0.01). This finding is consistent with previous results [21, 22].Fig. 2Distributions of serum neurofilament light chain (sNfL) concentrations and Ln-transformed distributions of sNfL levels Distributions of serum neurofilament light chain (sNfL) concentrations and Ln-transformed distributions of sNfL levels Associations of the WWI and ABSI with sNfL levels In the fully tuned model, an affirmative correlation was discerned between WWI and serum neurofilament light protein (sNfL) (β = 0.05, 95% CI 0.01–0.09). This suggested that the ln-transformed sNfL level increased by 0.05 units for every single increase in WWI. However, the ABSI showed a stronger positive correlation with ln-transformed sNfL levels in the completely corrected model (β = 1.65, 95% CI 3.53–13.72). This finding indicates that each incremental unit increase in ABSI correlates with a 1.65-unit elevation in the ln-transformed sNfL level. Furthermore, when the WWI and ABSI were further transformed from continuous to categorical variables (expressed in quartiles), greater quartiles of the WWI and ABSI were shown to correlate with higher ln-transformed sNfL levels than the lowest quartile category. The top percentile of WWI had a 0.13-unit greater ln-transformed sNfL level than did the lowest quartile (β = 0.13, 95% CI 0.05–0.22), whereas the ABSI’s highest quartile group had 0.17 units greater ln-transformed sNfL than did the lowest quartile (β = 0.17, 95% CI 0.08–0.27). As the WWI and ABSI increased, their positive associations with ln-transformed sNfL levels became more pronounced (P for trend < 0.01). Table 2 displays all the outcomes.Table 2The associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)ExposuresModel 1 β, (95% CI)Model 2 β, (95% CI)Model 3 β, (95% CI)WWI0.18(0.15, 0.22)0.04 (0.01,0.08)0.05 (0.01,0.09)Quartiles of WWIQ1 (8.89–10.42)ReferenceReferenceReferenceQ2 (10.43–10.97)0.09 (0.01,0.17)− 0.03 (− 0.10,0.04)− 0.01 (− 0.08,0.06)Q3 (10.96–11.56)0.20 (0.12,0.28)− 0.01 (− 0.08,0.07)0.02 (− 0.06,0.09)Q4 (11.57–14.11)0.39 (0.31,0.47)0.12 (0.04,0.20)0.13 (0.05,0.22)P for trend0.20 (0.16,0.24)0.06 (0.02,0.10)0.07 (0.03,0.11)ABSI2.99 (2.23,3.74)1.81 (0.92,2.70)1.65 (0.72,2.57)Quartiles of ABSIQ1 (0.29–0.38)ReferenceReferenceReferenceQ2 (0.38–0.41)0.06 (− 0.02,0.14)− 0.01 (− 0.08,0.06)− 0.01 (− 0.08,0.07)Q3 (0.41–0.44)0.15 (0.07,0.22)0.08 (− 0.00,0.16)0.09 (0.01,0.17)Q4 (0.44–0.56)0.30 (0.21,0.38)0.19 (0.09,0.28)0.17 (0.08,0.27)P for trend3.12 (2.27,3.98)2.06 (1.08,3.05)1.95 (0.94,2.96)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Table 2The associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)ExposuresModel 1 β, (95% CI)Model 2 β, (95% CI)Model 3 β, (95% CI)WWI0.18(0.15, 0.22)0.04 (0.01,0.08)0.05 (0.01,0.09)Quartiles of WWIQ1 (8.89–10.42)ReferenceReferenceReferenceQ2 (10.43–10.97)0.09 (0.01,0.17)− 0.03 (− 0.10,0.04)− 0.01 (− 0.08,0.06)Q3 (10.96–11.56)0.20 (0.12,0.28)− 0.01 (− 0.08,0.07)0.02 (− 0.06,0.09)Q4 (11.57–14.11)0.39 (0.31,0.47)0.12 (0.04,0.20)0.13 (0.05,0.22)P for trend0.20 (0.16,0.24)0.06 (0.02,0.10)0.07 (0.03,0.11)ABSI2.99 (2.23,3.74)1.81 (0.92,2.70)1.65 (0.72,2.57)Quartiles of ABSIQ1 (0.29–0.38)ReferenceReferenceReferenceQ2 (0.38–0.41)0.06 (− 0.02,0.14)− 0.01 (− 0.08,0.06)− 0.01 (− 0.08,0.07)Q3 (0.41–0.44)0.15 (0.07,0.22)0.08 (− 0.00,0.16)0.09 (0.01,0.17)Q4 (0.44–0.56)0.30 (0.21,0.38)0.19 (0.09,0.28)0.17 (0.08,0.27)P for trend3.12 (2.27,3.98)2.06 (1.08,3.05)1.95 (0.94,2.96)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted The associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml) *Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Results of subgroup analysis The long-term stability of the beneficial linking of WWI, ABSI, and ln-transformed sNfL levels across different cohorts were confirmed via subgroup studies. The differences in the correlation between WWI, ABSI, and ln-transformed sNfL levels across different strata were not statistically significant, indicating that this positive correlation does not significantly depend on sex, age, race, stroke, smoking, or congestive heart failure (P for interaction > 0.05). Specifically, among those under 60 years of age, a unit increase in WWI was linked to a 0.12-unit increase in logarithmically modified sNfL levels (β = 0.12, 95% CI = 0.08–0.16), whereas ABSI exhibited a 3.44-unit increase (β = 3.44, 95% CI 2.42–4.46). Among those 60 years of age and above, ABSI explained 3.16 units of a boost (β = 3.16, 95% CI 1.35–4.96), whereas a unit surge in WWI was linked with a 0.12-unit elevation in logarithmically converted sNfL levels (β = 0.12, 95% CI 0.05–0.22). However, age did not substantially impact the correlation among WWI, ABSI, and logarithmically transformed sNfL levels (WWI: P for interaction > 0.05; ABSI: P for interaction > 0.05). Tables 3 and 4 display all the outcomes.Table 3Subgroup regression analysis between the WWI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)CharacteristicModel 1 β(95%CI)Model 2 β(95%CI)Model 3 β(95%CI)P for interaction*Stratified by sex0.6502 Male0.23 (0.17, 0.28)0.04 (-0.02, 0.10)0.03 (− 0.03, 0.09) Female0.20 (0.15, 0.25)0.05 (0.01, 0.10)0.05 (0.00, 0.10)Stratified by age (years)0.7015 < 600.08 (0.05, 0.12)0.13 (0.09, 0.17)0.12 (0.08, 0.16) > = 600.13 (0.04, 0.22)0.15 (0.06, 0.23)0.14 (0.05, 0.22)Stratified by race0.1193 Mexican American0.13 (0.00, 0.26)− 0.00 (− 0.13, 0.13)− 0.02 (− 0.15, 0.11) Other Hispanic0.05 (− 0.11, 0.21)− 0.16 (− 0.33, 0.00)− 0.17 (− 0.34, − 0.01) Non-Hispanic White0.22 (0.18, 0.27)0.07 (0.03, 0.12)0.07 (0.02, 0.11) Non-Hispanic Black0.06 (− 0.04, 0.16)− 0.03 (− 0.13, 0.08)− 0.03 (− 0.14, 0.07) Other race—including multi-racial0.23 (0.10, 0.36)0.14 (− 0.00, 0.27)0.15 (0.01, 0.29)Stratified by stroke0.0607 Yes0.24 (-0.02, 0.49)0.27 (0.03, 0.51)0.26 (0.02, 0.51) No0.17 (0.13, 0.21)0.03 (− 0.00, 0.07)0.03 (− 0.01, 0.07)Stratified by smoking at least 100 cigarettes in life0.5200 Yes0.18 (0.12, 0.23)0.03 (− 0.02, 0.08)0.02 (− 0.03, 0.08) No0.03 (− 0.02, 0.08)0.05 (0.00, 0.09)0.05 (− 0.00, 0.09)Stratified by congestive heart failure0.4102 Yes0.10 (− 0.14, 0.34)0.11 (− 0.12, 0.34)0.13 (− 0.10, 0.36) No0.18 (0.14, 0.21)0.04 (0.00, 0.07)0.03 (− 0.00, 0.07)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjustedTable 4Subgroup regression analysis between the ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)CharacteristicModel 1 β(95%CI)Model 2 β(95%CI)Model 3 β(95%CI)P for interaction*Stratified by sex0.5414 Male5.84 (4.46, 7.23)2.37 (0.91, 3.83)2.12 (0.66, 3.59) Female4.79 (3.66, 5.92)1.61 (0.49, 2.74)1.55 (0.42, 2.68)Stratified by age (years)0.7838 < 600.95 (0.12, 1.79)3.70 (2.69, 4.72)3.44 (2.42, 4.46) > = 601.82 (0.27, 3.37)3.21 (1.39, 5.02)3.16 (1.35, 4.96)Stratified by race0.1552 Mexican American2.21 (− 0.22, 4.64)0.44 (− 2.38, 3.26)0.28 (− 2.54, 3.10) Other Hispanic3.33 (0.12, 6.54)0.15 (− 3.68, 3.98)0.03 (− 3.84, 3.91) Non-Hispanic White4.08 (3.13, 5.02)2.15 (1.05, 3.25)1.99 (0.89, 3.10) Non-Hispanic Black1.26 (− 0 .98, 3.50)2.21 (− 0.60, 5.02)2.09 (− 0.75, 4.92) Other race—including multi-racial2.62 (0.23, 5.01)2.53 (− 0.53, 5.59)2.53 (− 0.53, 5.59)Stratified by stroke0.7995 Yes1.92(-2.92,6.76)2.46(− 2.52,7.44)0.98 (− 4.33, 6.30) No2.81(2.05,3.57)1.75(0.85,2.66)1.68 (0.77, 2.58)Stratified by smoking at least 100 cigarettes in life0.5357 Yes2.94(1.73,4.14)1.50(0.17,2.83)1.35 (− 0.01, 2.70) No3.02(2.07,3.98)1.88 (0.80, 2.97)1.87 (0.77, 2.97)Stratified by congestive heart failure0.8051 Yes0.05 (− 5.53, 5.62)0.69 (− 6.03, 7.41)0.83 (− 5.86, 7.52) No2.93 (2.18, 3.69)1.81 (0.92, 2.71)1.67 (0.78, 2.57)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Table 3Subgroup regression analysis between the WWI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)CharacteristicModel 1 β(95%CI)Model 2 β(95%CI)Model 3 β(95%CI)P for interaction*Stratified by sex0.6502 Male0.23 (0.17, 0.28)0.04 (-0.02, 0.10)0.03 (− 0.03, 0.09) Female0.20 (0.15, 0.25)0.05 (0.01, 0.10)0.05 (0.00, 0.10)Stratified by age (years)0.7015 < 600.08 (0.05, 0.12)0.13 (0.09, 0.17)0.12 (0.08, 0.16) > = 600.13 (0.04, 0.22)0.15 (0.06, 0.23)0.14 (0.05, 0.22)Stratified by race0.1193 Mexican American0.13 (0.00, 0.26)− 0.00 (− 0.13, 0.13)− 0.02 (− 0.15, 0.11) Other Hispanic0.05 (− 0.11, 0.21)− 0.16 (− 0.33, 0.00)− 0.17 (− 0.34, − 0.01) Non-Hispanic White0.22 (0.18, 0.27)0.07 (0.03, 0.12)0.07 (0.02, 0.11) Non-Hispanic Black0.06 (− 0.04, 0.16)− 0.03 (− 0.13, 0.08)− 0.03 (− 0.14, 0.07) Other race—including multi-racial0.23 (0.10, 0.36)0.14 (− 0.00, 0.27)0.15 (0.01, 0.29)Stratified by stroke0.0607 Yes0.24 (-0.02, 0.49)0.27 (0.03, 0.51)0.26 (0.02, 0.51) No0.17 (0.13, 0.21)0.03 (− 0.00, 0.07)0.03 (− 0.01, 0.07)Stratified by smoking at least 100 cigarettes in life0.5200 Yes0.18 (0.12, 0.23)0.03 (− 0.02, 0.08)0.02 (− 0.03, 0.08) No0.03 (− 0.02, 0.08)0.05 (0.00, 0.09)0.05 (− 0.00, 0.09)Stratified by congestive heart failure0.4102 Yes0.10 (− 0.14, 0.34)0.11 (− 0.12, 0.34)0.13 (− 0.10, 0.36) No0.18 (0.14, 0.21)0.04 (0.00, 0.07)0.03 (− 0.00, 0.07)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Subgroup regression analysis between the WWI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml) *Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Table 4Subgroup regression analysis between the ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)CharacteristicModel 1 β(95%CI)Model 2 β(95%CI)Model 3 β(95%CI)P for interaction*Stratified by sex0.5414 Male5.84 (4.46, 7.23)2.37 (0.91, 3.83)2.12 (0.66, 3.59) Female4.79 (3.66, 5.92)1.61 (0.49, 2.74)1.55 (0.42, 2.68)Stratified by age (years)0.7838 < 600.95 (0.12, 1.79)3.70 (2.69, 4.72)3.44 (2.42, 4.46) > = 601.82 (0.27, 3.37)3.21 (1.39, 5.02)3.16 (1.35, 4.96)Stratified by race0.1552 Mexican American2.21 (− 0.22, 4.64)0.44 (− 2.38, 3.26)0.28 (− 2.54, 3.10) Other Hispanic3.33 (0.12, 6.54)0.15 (− 3.68, 3.98)0.03 (− 3.84, 3.91) Non-Hispanic White4.08 (3.13, 5.02)2.15 (1.05, 3.25)1.99 (0.89, 3.10) Non-Hispanic Black1.26 (− 0 .98, 3.50)2.21 (− 0.60, 5.02)2.09 (− 0.75, 4.92) Other race—including multi-racial2.62 (0.23, 5.01)2.53 (− 0.53, 5.59)2.53 (− 0.53, 5.59)Stratified by stroke0.7995 Yes1.92(-2.92,6.76)2.46(− 2.52,7.44)0.98 (− 4.33, 6.30) No2.81(2.05,3.57)1.75(0.85,2.66)1.68 (0.77, 2.58)Stratified by smoking at least 100 cigarettes in life0.5357 Yes2.94(1.73,4.14)1.50(0.17,2.83)1.35 (− 0.01, 2.70) No3.02(2.07,3.98)1.88 (0.80, 2.97)1.87 (0.77, 2.97)Stratified by congestive heart failure0.8051 Yes0.05 (− 5.53, 5.62)0.69 (− 6.03, 7.41)0.83 (− 5.86, 7.52) No2.93 (2.18, 3.69)1.81 (0.92, 2.71)1.67 (0.78, 2.57)*Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Subgroup regression analysis between the ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml) *Model 1 = no covariates were adjusted. Model 2 = Model 1 + age, sex and race were adjusted. Model 3 = all covariates were adjusted Validation of the linear relationship of WWI and ABSI with sNfl levels To evaluate the potential nonlinear interactions among the predictor variables and ln-transformed sNfL levels, smoothing curve fitting and threshold effect analysis were performed. There was a curvilinear correlation between WWI and ln-transformed sNfL levels in Fig. 3. The threshold effect analysis subsequently revealed a negative correlation prior to the inflection value of 10.38 and a positive relationship thereafter (LLR < 0.001). Figure 4 shows that a nonlinear correlation with the ABSI and ln-transformed sNfL when the ABSI was treated as the independent factor. This nonlinear relationship was further validated through a threshold effect study (LLR = 0.002), as shown in Table 5.Fig. 3The association between the WWI and sNfL ln-transformed (pg/ml). The solid line represents the smoothing curve fitting between the variables. The blue band indicates the fitted 95% confidence intervalFig. 4The association between the ABSI and sNfL ln-transformed (pg/ml). The solid line represents the smoothing curve fitting between the variables. The blue band indicates the fitted 95% confidence intervalTable 5Threshold effect analysis for associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)ExposuresWWIABSILinear effect modelβ, (95%CI)0.05 (0.01, 0.09)1.65 (0.72, 2.57)Non-linear modelInflection point (K)10.380.38β, (95%CI) (< K)− 0.16 (− 0.28, − 0.05)− 2.06 (− 4.61, 0.49)β, (95%CI) (≥ K)0.10 (0.05, 0.15)2.54 (1.45, 3.63)LLR < 0.0010.002*All covariates in additional were adjusted The association between the WWI and sNfL ln-transformed (pg/ml). The solid line represents the smoothing curve fitting between the variables. The blue band indicates the fitted 95% confidence interval The association between the ABSI and sNfL ln-transformed (pg/ml). The solid line represents the smoothing curve fitting between the variables. The blue band indicates the fitted 95% confidence interval Table 5Threshold effect analysis for associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml)ExposuresWWIABSILinear effect modelβ, (95%CI)0.05 (0.01, 0.09)1.65 (0.72, 2.57)Non-linear modelInflection point (K)10.380.38β, (95%CI) (< K)− 0.16 (− 0.28, − 0.05)− 2.06 (− 4.61, 0.49)β, (95%CI) (≥ K)0.10 (0.05, 0.15)2.54 (1.45, 3.63)LLR < 0.0010.002*All covariates in additional were adjusted Threshold effect analysis for associations of the WWI and ABSI with ln-transformed serum neurofilament light chain (sNfL) levels (pg/ml) *All covariates in additional were adjusted Discussion Our cross-sectional survey revealed significant positive associations between WWI and ABSI and between WWI and ABSI levels. Subgroup analyses and interaction evaluation revealed that this positive association remained consistent across all categories. Notably, we found nonlinear associations between the WWI and ABSI and sNfL, with inflection points of 10.38 and 0.38, respectively. To our knowledge, this work is the first to investigate the associations among WWI, ABSI and sNfL, providing preliminary evidence for the impact of obesity on neurodegenerative diseases. What is already known on this subject? Reviewing previous studies, various cross-sectional studies have explained the associations between neural and nonneural factors and sNfL. Neurodegenerative diseases or traumatic impacts can cause neuronal damage, leading to increased sNfL levels [23]. Among nonneural factors, models predicting sNfL concentrations have shown significantly increased explanatory power when age, cardiovascular factors, and kidney function are considered[24]. Recent research has indicated that increases in sNfL are associated with environmental exposure to glyphosate (an herbicide), phthalates (plastic additives), and smoke[17, 25, 26]. Given recent research on sNfL, a more comprehensive and extensive exploration of this advanced biomarker for neurodegenerative changes may aid in clinical decision-making and treatment. What this study adds? Our data are the first to demonstrate a significant positive association between the new adiposity index and sNfL, with a potential nonlinear association. We noted that this negative correlation before the inflection point was consistent with the previous literature showing that obesity was inversely associated with sNfL[21, 27]. This may be related to the dilution effect, where a higher BMI may lead to greater blood volume to dilute the concentration of Nfl, resulting in lower Nfl levels in the serum. The positive correlation after the inflection point suggested that excessive obesity leads to comorbidities such as diabetes, hypertension, and stroke. These diseases may cause damage to nerve axons through ischaemia, metabolic abnormalities, inflammation, and oxidative stress, which in turn increases the concentration of sNfl. Korley et al. reported that stroke, hypertension, poor glycaemic control, and cardiovascular risk factors, including renal function, were also associated with increased sNfL [28]. Therefore, the nonlinear relationship between the WWI and ABSI with sNfl indicated that not all adiposity measures were inversely associated with the sNfL concentrations determined thus far. Furthermore, Korley et al. reported that sNfl may be cleared by the kidney, so renal function should be considered when interpreting sNfl levels [28]. As presented in our data, a high eGFR corresponds to a low sNfl, which is consistent with previous data in the literature. Eleni Rebelos et al. reported that Nfl levels increased after weight loss induced by bariatric surgery and that Nfl was negatively associated with eGFR both at baseline and after bariatric surgery, suggesting that the extent of renal function may influence circulating Nfl levels [22]. They proposed that the decreased circulating NfL in obese patients may be based on their improved glomerular filtration rate. Importantly, a key feature of obesity is glomerular hyperfiltration [29]. This is a compensatory mechanism, but in the long term, it forms the basis of obesity-induced chronic renal failure [30]. This also explains the inflection point in our data, before which obesity-induced glomeruli are considered compensatory patterns affecting elevated sNfl levels and, to a greater degree, obesity-induced chronic renal failure triggers elevated sNfl levels. In turn, this confirms the contributions of Eleni Rebelos et al. Since obesity is associated with reduced brain density [31], it represents a risk factor for neurodegeneration. Moderate and chronic inflammation is often linked to obesity [32, 33]. The body's production of inflammatory mediators [34] has been verified to be positively correlated with NfL and inflammatory cytokines (IL-1β and IL-6) [35]. This finding aligns with prior animal studies indicating that elevated levels of inflammatory mediators (IL-2, IL-4, IL-6, and TNF-α) and increased CSF NfL levels are linked to chronic neuroinflammation and immune responses [36]. This finding further corroborates the potential mechanisms linking obesity with sNfL levels. sNfL has been proposed as a valuable indicator for various neurological disorders, especially for predicting cognitive decline and the severity of brain lesions [37]. Recent progress in elucidating the molecular mechanisms of obesity and neurodegenerative diseases has identified astrocytes and endoplasmic reticulum stress as crucial links bridging obesity and neurodegenerative pathologies [38]. Current advancements suggest that neuroinflammation serves as an alternative route connecting obesity and Alzheimer's disease [39]. This increased the possibility of using innovative clinical approaches to treat and prevent neurodegenerative illnesses. Therefore, this research suggests that the correlation between obesity and sNfL provides guidance for clinical interventions in high-risk populations for neurodegenerative diseases. Although the specific mechanisms linking obesity and neurodegenerative diseases require further investigation, effective weight management, maintaining physical health, and regular health check-ups provide valuable insights for clinical guidelines. Strengths and limitations NHANES, a broadly representative cross-sectional survey, is our study's primary strength. Sampling weights facilitate the evaluation of WWI, ABSI, and sNfL relationships. To ensure study accuracy, the authors controlled for confounding factors. Subgroup analyses were also performed to determine the reliability of the WWI, ABSI, and sNfL correlations across populations. To assess any nonlinear correlations, we also performed threshold effect analysis and smoothed curve fitting. However, we acknowledge the limitations of our investigation. First, causal inferences cannot be made based on our cross-sectional study despite strong correlations among WWI, ABSI, and sNfL levels. Therefore, further validation is needed in prospective cohort studies. Second, the sNfL data utilized in this study were confined to the NHANES 2013–2014 cycle, where sNfL measurements were available. Given the absence of sNfL data in other NHANES cycles, our findings may not fully represent the cross-sectional profiles of these patients. Consequently, further large-scale longitudinal cohort studies are warranted to validate our initial observations. Moreover, although we accounted for numerous covariates, we recognize the potential for biases stemming from unmeasured confounding variables. These factors may involve drug use and long-term illness, including diabetes, renal disease, and hepatic disorders, among others. Finally, despite the use of rigorous methodological approaches, our study is hampered by the relative population size for subgroup estimation. To yield more dependable and definitive outcomes in future experimental subgroups, expanding the sample size significantly is imperative.