There has been a growing body of literature discussing the link between obesity and inflammation. Obesity has been described as a state of low-grade, chronic inflammation. This inflammatory state is initiated by metabolic triggers like excess nutrition and mediated by metabolically-active cells such as adipocytes capable of producing cytokines, including TNF-α. Production of these cytokines leads to the activation of the inflammatory cascade [1-4].

Given the well-established relationship between autoimmunity and inflammation, the association of obesity with inflammation has led to the question of a potential link between obesity and autoimmunity [5,6]. Recently Crowson et al. have shown an association between obesity and the development of rheumatoid arthritis, with inflammation as the common factor between the two conditions [7]. A relationship between obesity and autoimmunity has also been found in multiple other conditions [8]. Given that anti-nuclear antibodies (ANAs) are correlated and are also pathologic in multiple autoimmune and rheumatic diseases, steps have been taken to look for the presence of these auto-antibodies in obese patients [9,10].

Satoh et al. examined key correlates, including obesity, that contribute to the prevalence of ANA in the general population through a cross-sectional analysis of individuals from the National Health and Nutrition Examination Survey 1999-2004 (NHANES) [11]. They found that ANAs were less common in overweight and obese individuals, a seemingly counterintuitive finding as one would expect increased levels of autoimmune markers related to inflammation in this group. Gonzalez et al. have also previously described an inverse relationship between ANA and obesity in women [12].

Due to these findings, we sought to explore the interplay of obesity, inflammation and autoimmunity in the general population. We examined the NHANES cohort to further investigate the association between ANA and obesity as well as to identify key mediators in this relationship. In order to uncover previously unidentified factors in this relationship, obesity and body composition were examined using other measures beyond the Body-Mass Index (BMI), including measures of skeletal muscle mass and the distribution of body fat.

Study population

NHANES 1999-2004 was a nationally representative survey of the non-institutionalized civilian population in the United States [13]. A stratified, multistage, probability sampling design was used to select participants. The NHANES protocol was approved by the National Center for Health Statistics ethics review board and written informed consent was obtained from all participants. Overall, 3,564 adults aged ≥ 20 years completed the interview and examination components and had ANA measurements. In order to exclude individuals with possible autoimmune disease, we excluded participants who reported diagnoses of rheumatoid arthritis or arthritis other than osteoarthritis (n=278); steroid use within the previous month (n=39); thyroid disease or the use of thyroid medication (n=274); or liver disease (n=92). After excluding participants with missing body composition or other covariate data (n=329), 2,552 participants were available for analysis.

Data collection

Information on education, smoking status, and comorbidities was obtained by self-report. Race/ethnicity was self-identified. Smoking was classified as never, former, or current smoker. Hypertension was defined as a systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, physician diagnosis, and/or antihypertensive medication use [14]. A participant was considered to have diabetes mellitus if he or she reported a physician diagnosis while not pregnant or the current use of insulin or oral hypoglycemic medications, or had a glycohemoglobin level ≥ 6.5%. Cardiovascular Disease (CVD) was defined by self-report of a physician diagnosis of congestive heart failure, coronary heart disease, angina, myocardial infarction, or stroke.

Body-Mass Index (BMI, kg/m²) was categorized as underweight (<18), normal weight (18-24.9), overweight (25-29.9), and obese (≥ 30). Waist Circumference (WC) was available for 1,644 participants and was used to define abdominal obesity as WC ≥ 88 cm for women and ≥ 102 cm for men [15]. Total body fat, trunk fat, and Lean Body Mass (LBM) were assessed using whole-body Dual-Energy X-Ray Absorptiometry (DXA). Due to the pattern of non-response, missing and invalid data were multiply imputed. Details of the DXA protocol, quality control analyses, and the multiple imputation procedure are available (http://www.cdc.gov/nchs/nhanes/dxx/dxa.htm). Percent total body fat was calculated as 100 × total body fat/total mass, and % trunk and non-trunk fat were calculated as 100 × trunk fat/trunk mass and 100 × non-trunk fat/non-trunk mass, respectively. Truncal body fat ratio was calculated as trunk fat/non-trunk fat. Appendicular Skeletal Muscle Mass Index (ASMI) was calculated as the sum of lean mass for the arms and legs (kg)/height² (m²) [16]. ASMI is a measure of relative muscle mass that is analogous to BMI as a measure of relative adiposity.

Total cholesterol and triglycerides were measured using coupled enzymatic reactions, HDL cholesterol was measured using a heparin-manganese precipitation method, and LDL cholesterol was calculated from these values. C - reactive protein (CRP) was quantified by latex-enhanced nephelometry. High and low CRP were defined using the 75^th percentile cutpoint in our cohort as ≥ 0.42 and <0.42 mg/dL, respectively. Fasting glucose and insulin were used to calculate homeostasis model assessment of insulin resistance (HOMA-IR), a model for estimating insulin resistance [17]. Glucose was measured using the glucose hexokinase method and insulin was measured by radioimmunoassay. HOMA-IR ≥ 2.5 was categorized as insulin resistant. Serum triglycerides, HDL and LDL cholesterol, and HOMAIR were examined in the 730 participants with available data for each of these parameters.

Outcome variables

ANA testing was performed in a subsample of participants from 1999-2004 using the HEp-2 cell immunofluorescence assay (INOVA Diagnostics, San Diego, CA). In samples with a ≥ 3+ nuclear and/or cytoplasmic immunofluorescence pattern, ANA titers were determined by serial dilution. Full details of the protocol are available at http:// www.cdc.gov/nchs/nhanes/nhanes2003-2004/SSANA_C.html. We selected a stringent and highly specific definition of ANA positivity, defining a positive ANA as a titer ≥ 1:160.

Statistical analysis

All analyses used NHANES-appropriate sampling weights and accounted for the complex multistage cluster design using the “survey” command in Stata 11.2 (Stata Corporation, College Station, TX, USA). Specifically, the svyset command was used to specify the survey design characteristics, and all estimation commands were preceded by the svy prefix. A p-value <0.05 was considered statistically significant. The distributions of participant characteristics were examined by categories of BMI. Logistic regression models were created to examine the association of BMI categories with ANA positivity, first without adjustment for additional covariates and then including age, sex, and race/ethnicity as covariates. Multivariable models were then created including variables determined a priori to be potential confounders of the association of BMI with ANA positivity, including smoking status, education (as a marker of socioeconomic status), diagnosis of diabetes mellitus, hypertension, CVD, and total cholesterol. Finally, log-transformed CRP was added to the multivariable models to separately examine the effect of systemic inflammation.

We tested the possibility of effect modification by inflammation of the association of BMI with ANA positivity by first including a multiplicative interaction term of log-transformed CRP and BMI categories in the multivariable model. We determined a priori that a p-value <0.20 would be considered suggestive of interaction. We then examined associations of BMI categories with ANA positivity in the high and low CRP subgroups, separately. Based on these results, we next examined the distributions of participant characteristics including body composition by ANA status in overweight and obese participants only, and in the overweight and obese subgroup by high and low CRP status, separately. We further explored the importance of body composition by categorizing participants into sex-specific quartiles of truncal body fat ratio and examining the association with ANA positivity in the high and low CRP subgroups, separately.

Of the 2552 participants, 933 were overweight and 743 were obese (Table 1).

Table 1: Participant characteristics by categories of Body-Mass Index.

Participants with higher BMI were older (p<0.001) and more likely to be men (p=0.004). Overall, participants in the higher weight categories had higher rates of hypertension (p<0.001), diabetes (p<0.001) and cardiovascular disease (p<0.001). Interestingly, while the overweight and obese participants were less likely to be ANA positive compared to those with normal BMI (p=0.01), they were substantially more likely to have a high CRP level (p<0.001). For further results please refer to Table 1.

Using a simple logistic regression model we confirmed the result found by Satoh et al. that obese individuals are less likely to be ANA positive [11]. Obesity was associated with a significantly decreased risk of ANA positivity (OR: 0.74, 95%CI (0.59-0.93)). This association remained after adjustment for multiple covariates including age, sex, race/ethnicity and multiple comorbidities as detailed in (Table 2).

Table 2: Odds ratio of positive antinuclear antibodies by body-mass index in 2,552 participants of NHANES 1999-2004.

However the association between ANA positivity and obesity was no longer significant after adjustment for log-transformed CRP (OR: 0.85, 95%CI (0.62-1.15)). Furthermore, the p-value for interaction (p=0.12) by CRP was suggestive of an effect modification in the association of BMI with ANA positivity.

To further evaluate the role of inflammation in the association between obesity and ANA positivity, we stratified our logistic regression models based on CRP level (Table 3).

Table 3: Odds ratio of positive antinuclear antibodies by body-mass index and C-reactive protein concentration in 2,552 participants of NHANES 1999-2004.

Among the 1676 overweight and obese participants (Table 4), those who were ANA positive were more likely to be female (p<0.001).

Table 4: Characteristics by antinuclear antibody status in 1,676 overweight and obese participants of NHANES 1999-2004.

Otherwise, ANA positive and ANA negative participants were similar in terms of demographic characteristics and comorbidities, with the exception of a higher prevalence of CVD in the ANA positive group (12.6% vs. 7.5%, p=0.02). However, overweight and obese ANA positive and negative participants differed in terms of body habitus. Although the groups were comparable in terms of BMI, there was a trend toward higher waist circumference in the ANA positive group, and percentages of total body fat (p=0.007), trunk fat (p=0.02) and non-trunk fat (p=0.004) were each higher among ANA positive versus ANA negative participants. In addition, ANA positive participants had lower ASMI compared to the ANA negative group (7.88 vs. 8.35, p<0.001).

Among obese and overweight participants with high CRP, there was no significant difference in fat distribution between the ANA positive and ANA negative participants (Table 5).

Table 5: Characteristics by antinuclear antibody status and C-reactive protein concentration in 1,676 overweight and obese participants of NHANES 1999-2004.

However, in those with low CRP, there was a significant difference in body habitus between the ANA groups. ANA positive participants had higher percentages of abdominal obesity (p=0.04), total body fat (p=0.006) trunk fat (p=0.009), and non-trunk fat (p=0.005). In addition to the differences seen in fat distribution between the groups, the ASMI was significantly lower in the ANA positive patients regardless of their CRP level.

Given the association of truncal fat with ANA positivity, we repeated the analyses presented in Table 3 using quartiles of truncal body fat ratio instead of BMI categories. With the inclusion of this variable into the models, the association between ANA positivity and obesity is no longer statistically significant in those with a low CRP. Therefore the relationship seen between ANA positivity and obesity is in fact partially explained by the presence of truncal obesity (Table 6).

Table 6: Odds ratio of positive antinuclear antibodies by truncal body fat ratio and C-reactive protein concentration in 2,552 participants of NHANES 1999-2004.

We have found that the relationship between obesity and autoimmunity in the general population is partly mediated by systemic inflammation. We first confirmed the previously reported, yet seemingly counterintuitive association of obesity with a lesser likelihood of ANA positivity. However, this association was no longer significant after accounting for levels of CRP. The relationship of obesity with autoimmunity was modified by the presence of systemic inflammation, as measured by serum CRP. Specifically, among participants in the highest quartile of CRP, being overweight or obese was associated with an increased risk of ANA positivity. Although not statistically significant, this trend was in sharp contradistinction to our results among participants with lower levels of serum CRP, in whom the opposite association was found. Furthermore, among overweight and obese participants with low serum CRP, ANA positivity was associated with a greater prevalence of not only cardiovascular disease and hypertension but also higher percentages of total body fat and abdominal obesity. These data suggest an important interplay of inflammation, body composition, and autoimmunity in the general population.

It is a well-acknowledged that obesity is a chronic inflammatory state where this inflammation plays a key role in the metabolic syndrome and in the relationship between obesity and cardiovascular disease. While general obesity is associated with the metabolic syndrome, abdominal/visceral adiposity appears to be the driving force behind this association [1,18]. Within obese populations there are two major constituents, those with high rates of the metabolic disturbances that constitute the metabolic syndrome– hypertension, hyperlipidemia and insulin resistance-, and those “healthy obese” without such derangements. A major difference between the two groups is body habitus where those with the metabolic syndrome have higher rates of abdominal adiposity [19]. Central adiposity has been shown to be associated with higher rates of C-reactive protein (CRP), a common measure of systemic inflammation, in all age groups as well as in people with normal BMI who have higher abdominal fat deposits [20-22].

As has been previously reported, in our study obese participants had elevated levels of inflammation as measured by CRP at baseline [20,21]. Therefore when evaluating the association between obesity and ANA positivity it is imperative to consider the role of inflammation given that it is typically inflammatory conditions such as lupus that induce ANA positivity. As to be expected, in those obese participants with CRP levels higher than the median, obesity was associated with an increased risk of being ANA positive. While the result was not statistically significant it is possible that this is due to the small sample size in this group. However we see that in the low CRP group there is still decreased risk of being ANA positive if obese.

In order to determine what could potentially be driving this relationship we evaluated the individual characteristics of the ANA positive and negative participants within the overweight and obese as stratified by CRP levels. Overall in the high CRP group there was no difference between the ANA positive and negative participants with regards to the presence of various co-morbidities or body habitus. However in the low CRP group the ANA positive participants were more likely to be hypertensive, have cardiovascular disease and have higher percentages not only of total body fat and truncal obesity. Therefore we postulate that even in the setting of low systemic inflammation, these ANA positive, overweight and obese participants are likely in an inflammatory state that is causing them to be ANA positive.

It has been shown through animal models as well has human samples that Visceral Adipose Tissue (VAT) is metabolically active and produces inflammation through various cytokines, chemokines, and adipokines [1,2]. These molecules, which include TNF-alpha, IL-6, IL-8, and MCP-1, activate other inflammatory mediators including macrophages, T cells and B cells, and other adipocytes to further the inflammatory cascade [22-25]. Activated adipose tissue may then play a role in the development of autoimmunity through the hormones leptin and adiponectin [26-28]. As an inflammatory hormone, leptin promotes T-cell activation and B-cell class switching, a step that is important in antibody development [29,30]. Adiponectin, an anti-inflammatory hormone, promotes the release of anti-inflammatory cytokines such as IL-1 and IL-10 but is inhibited by TNF-alpha and IL-6 [31,32]. Leptin also inhibits the apoptosis of antigen-specific T cells, while adiponectin facilitates the clearance of apoptotic bodies. Treating auto-immune prone mince with adiponectin decreases ANA levels. [31] Therefore, within the inflammatory milieu of obesity, low levels of adiponectin are likely insufficient to blunt the inflammatory response to leptin. Hence, increased cellular debris from inflammation could act as antigenic targets leading to the production of ANAs by B cells, possibly within the VAT itself. Therefore, obesity, through inflammation, could lead to autoimmunity [32].

Environmental factors, such as nutrition, are associated with multiple epigenetic changes. It has been shown that the intake of a diet consisting of high amounts of saturated fats can lead to significant epigenetic changes in human adipose tissue where these changes are potentially genome wide and are influenced by weightloss [33-35]. These epigenetic chromosomal changes may also affect the generation of autoantibodies in the setting of obesity. It has been shown in both lupus and Sjogern’s syndrome, two conditions characterized by the presence of ANAs and other auto-antibodies, that there are potentially pathologic epigenetic changes that are associated with auto-antibody production [36,37]. Given that epigenetic changes are implicated in the generation of autoimmunity, and that obesity has been linked to multiple epigenetic mechanisms, it is possible that nutritionally induced epigenetic changes are also driving autoimmunity in obese patients [38]. This could in fact potentially amplify whatever role, adipokines may play in autoimmunity. While the above proposed mechanisms are interesting, they remain speculative at this time.

If the ANA positive patients are part of the obese population in an inflammatory state that may potentially develop the metabolic syndrome, especially given their higher rates of hypertension and cardiovascular disease, those with low CRPs who are ANA negative may constitute a population of patients known as the “Metabolically Healthy but Obese” (MHO). MHO patients constitute approximately 20% of the obese population [29]. Overall the MHO population has similar BMIs however they tend to have lower amounts of visceral fat when compared to the obese at risk for the metabolic syndrome. Not only do MHOs have a different body habitus but these patients also have higher rates of insulin sensitivity, have high lean body mass and better lipid profiles [30]. This improved metabolic profile in the MHOs maybe secondary to the fact that they also have improved inflammatory profiles with lower levels of CRP and higher levels of adiponectin at baseline [32,33,39]. Because of their lower percentage of VAT, lower CRP and higher adiponectin levels, MHO patients are likely protected from developing the metabolic syndrome in the future.

Absent autoimmunity and relatively little systemic inflammation may therefore represent a new MHO subgroup. These characteristics were associated with lesser body fat and greater skeletal muscle mass among overweight and obese participants, despite no apparent difference in BMI. In contrast, among overweight and obese participants with high levels of CRP, there was no difference in adiposity based on ANA status. However, ANA positivity was still associated with lesser skeletal muscle mass and with a trend toward lower BMI. Taken together, these results suggest an important role of skeletal muscle in the interplay of obesity, inflammation, and autoimmunity that deserves further exploration. Indeed, a growing body of evidence suggests that skeletal muscle is a key regulator of metabolism, and that loss of muscle mass and impaired skeletal muscle function directly contribute to metabolic dysregulation and inflammation [40,41].

There are several potential limitations to our study. First, as this was a cross-sectional analysis, causality cannot be inferred. Also, we are unable to determine whether ANA positivity predicts future development of the metabolic syndrome or of a connective tissue disease. Our ascertainment of the components of the metabolic syndrome was limited as most participants did not have measurements of insulin resistance. It is also possible that ANA positive participants had an undiagnosed connective tissue disease. However this would be unlikely given that we excluded participants who reported having arthritis that was not osteoarthritis, those with a thyroid condition or liver disease, and anyone reporting recent steroid use. Nevertheless there is a remote possibility. Lastly, the only inflammatory marker available to us was CRP, we do not know baseline leptin or adiponectin levels, nor do we have access to samples from these participants VATs to then better associate these levels with the presence or absence of ANAs.

Despite the limitations of this study we feel that this data adds to the body of literature regarding the association of obesity with inflammation. This data demonstrates that there could potentially be another mechanism for inflammation in humans where there in not only inflammatory milieu created but B cells are activated and are potentially generating ANA locally within the VAT. We feel that this exciting data generates many new avenues of inquiry into inflammation in humans with visceral adiposity.