Sex differences in the association between antinuclear antibody positivity with diabetes and multimorbidity in older adults: results from the Baltimore Longitudinal Study of Aging

Abstract

Antinuclear antibodies (ANA), a marker of self-reactivity to DNA and other nuclear antigens, are present in several autoimmune diseases and have been observed in healthy persons in the absence of autoimmune disease. ANA prevalence is higher in women and older adults, but the health implications of ANA in middle- to older-aged adults are unknown. Immune system differences by sex may further result in sex-specific susceptibility to morbidity. In a cross-sectional analysis of data from the Baltimore Longitudinal Study of Aging, we examined the sex-specific relationship between age and ANA as well as the associations (odds ratios and 95% confidence intervals) between ANA and type-2 diabetes and multimorbidity (2 or more chronic diseases), stratified by sex and controlling for age and race. ANA was measured in a 1:160 dilution of sera by immunofluorescence using HEp-2 cells (seropositive = 3 or 4). Overall ANA seroprevalence was 12% (15.1% in women, 8.8% in men). We observed a non-linear relationship between age and ANA that varied by sex (interaction p-value <0.05), with a clear sex differences in younger participants (ages 48–59), which converged in the oldest (age 80+). ANA positive women had higher odds of type 2 diabetes (OR: 2.06, 95% confidence interval: 1.04, 4.07) and multimorbidity (OR: 2.47, 95% confidence interval 1.11, 5.50) than women who were ANA negative. No statistically significant associations were observed in men. Insight into differences in age-related ANA positivity and ANA associations with chronic diseases by sex is important for understanding the impact of immune dysregulation in aging individuals.

1. Introduction

Elevated antinuclear antibodies (ANA), autoantibodies to DNA and other antigens in the cell nucleus are a marker of self-reactivity seen across multiple autoimmune diseases^1,2 such as systemic lupus erythematosus, and may precede the development of autoimmune disease by several years.^3,4 Importantly, ANA have also been observed in persons without autoimmune disease^5,6 and ANA are present in 12–15% of the general U.S. population, with a higher prevalence in women and older adults.⁷ Both innate and adaptive immunological factors vary by sex across the life course resulting in sex-specific susceptibility to autoimmune diseases, cancer and infectious diseases.⁸ Though alterations in immune function that accompany the aging process may differ by sex⁹, few studies have explicitly examined sex differences in ANA positivity by age.

The incidence of most autoimmune diseases is low in older adults¹⁰ and in the absence of clinical autoimmune disease diagnoses, the health implications of ANA positivity in middle and older adults is not well known. Insight into sex differences in age-related ANA positivity, as well associations between ANA and morbidities, may improve understanding of the pathophysiologic pathways through which ANA positivity may be clinically meaningful in middle-aged and older adults.¹¹ Some morbidities, such as type-2 diabetes, have multiple phenotypes of which some are hypothesized to potentially have an autoimmune etiology.¹² Additionally, low grade inflammation resulting from continuous stimulation of innate and adaptive immune processes associated with aging may play an important role in immune dysregulation leading to the presence and the development of chronic diseases.^13,14 Addressing a key gap in knowledge, we examined sex-specific ANA positivity by age as well as the association between ANA and 1) diabetes and 2) multimorbidity using data from the Baltimore Longitudinal Study of Aging (BLSA).

2. Material and Methods

2.1. Study Population

The BLSA is a continuous enrollment cohort study of normative aging sponsored and conducted by the Intramural Research Program of the National Institute on Aging (NIA). The study was approved by the National Institute of Environmental Health Sciences Institutional Review Board. All participants gave written informed consent in accordance with the Declaration of Helsinki. Details of the BLSA have been described previously.¹⁵ Briefly, general BLSA enrollment is restricted to individuals free of cognitive impairment, functional limitations and chronic diseases except for controlled hypertension. Participants undergo a comprehensive health history and physical examination, collection of biological samples including blood draw and assessment of physical function at the NIA Clinical Research Unit. Visits occur every 4 years for individuals under age 60, biennially for individuals aged 60–79, and annually for individuals aged 80 and older.

For the present study, we conducted a retrospective cross-sectional analysis in a subsample of the main BLSA cohort who had ANA testing completed at one study visit between June 2000 and September 2014, resulting in a eligible study population consisting of 878 participants aged 48 to 103. Of the 878, persons with a diagnosis of rheumatoid arthritis or SLE or who used methotrexate, hydroxychloroquine or other disease-modifying antirheumatic drugs (DMARDs, for full list see Appendix A) were considered to have suspected autoimmune disease. As shown in Supplemental Table 1, these individuals (N= 15, 2%) had higher odds of ANA positivity than the study sample and were excluded from analyses focused on associations between ANA positivity and non-autoimmune disease morbidity. Additionally, individuals missing data for either diabetes or multimorbidity were excluded from those specific analyses, resulting in a final analytic sample of 721 for diabetes and 708 for multimorbidity.

2.2. Outcomes

Diabetes prevalence was measured by self-report of diagnosis by a medical professional of Type 2 diabetes, high blood sugar or glucose intolerance and categorized as “yes” or “no.” Multimorbidity is defined as the co-existence of 2 or more chronic diseases¹⁶, and there is heterogeneity in the chronic conditions contributing to multimorbidity indices. Most rely on self-reported information and focus on diseases with a high prevalence and a severe impact on affected individuals.¹⁶ We used self-reported health data on the most prevalent non-communicable diseases in our sample to generate a dichotomous multimorbidity variable (2 or more chronic conditions v. 1 or 0 chronic conditions). Other less common conditions were not included in our analysis due to low prevalence; for example, only 9 of 743 individuals (1.21%) had a COPD diagnosis, and several other health conditions sometimes included in multimorbidity indexes were also of low prevalence in our sample (cognitive impairment N=1, hip fracture N= 9, Parkinson’s N=7) and thus were not included in our measurement of multimorbidity. A dichotomous (2 or more morbidities versus 0 or 1 morbidity) multimorbidity variable was generated from the sum of the most prevalent conditions of interest, including reported diagnoses by a medical professional of diabetes (Type 2, high blood sugar or glucose intolerance), kidney disease, transient ischemic attack (TIA) or stroke, congestive heart failure, myocardial infarction, and cancer (excluding non-melanoma skin cancer).

2.3. Exposure

Stored serum samples were tested for immunoglobulin G (IgG) autoantibodies to human nuclear antigens using indirect immunofluorescent methods at the University of Colorado-Denver Clinical Research Laboratory under College of American Pathology/Clinical Laboratory Improvement Amendment (CAP/CLIA) certified conditions. HEp-2 cell slides (Kallestad, Bio-Rad Laboratories, Hercules, CA) were incubated with a 1:160 dilution of sera and then washed and incubated with FITC anti-human immunoglobulin reagent. Using fluorescent microscopy, trained technicians scored ANA fluorescence for each individual on a 0–4 scale. The distribution of ANA intensity by age and sex are presented in Supplemental Table 2. Individuals with ANA intensities of 3 or 4 at the 1:160 dilution were considered seropositive for ANA and evaluated in comparison to those with intensities of 0–1, who were classified as seronegative. This positivity classification system was developed by laboratory-specific experiments and a level that was >95% sensitive for patients with SLE.¹⁷ Participants with 2+ values were dropped from the analysis sample based on quality control data in subsample of participants re-tested for ANA fluorescence showing higher concordance for specimens that were either positive (3 or 4+) or negative (0 or 1+) for (91% concordance, excluding borderlines)original immunofluorescent assay results. Sensitivity analyses were conducted including individuals with borderline values as either positive or negative.

2.4. Covariates

Potential confounders were included as covariates in the statistical analysis based on a priori knowledge and associations with ANA in the study sample. Covariates in all analyses included age (years), sex (women referent) and self-identified Black or non-Black (referent) race. Previous work has shown that the relationship between age and ANA prevalence may be nonlinear⁷, therefore, continuous age was modeled using sex-specific restricted cubic splines to control for a non-linear relationship after examining the age-ANA distribution in the BLSA with knots at age 55, 68, 78 and 92 years of age corresponding to the 5^th, 35^th, 65^th and 95^th percentile of the age distribution. Body mass index (BMI, kg/m²) was categorized as <25kg/m², 25–30kg/m², or >30kg/m² and included as a covariate in the diabetes analysis due to known associations between BMI, ANA and diabetes. Among post-menopausal women, use of estrogen or progesterone hormone therapy (HT) (categorized as never, past or current) was not statistically significantly associated with ANA positivity and thus not included as a covariate in models.

2.5. Statistical Analysis

ANA was first characterized by categorical age and sex in the full sample. Heterogeneity in the age distribution of ANA was found by sex and formally tested by comparing nested logistic regression models using a likelihood ratio test. Descriptive statistics of ANA frequency by demographic characteristics in the total and sex-stratified samples were performed; age-adjusted associations between ANA and demographic characteristics were measured using logistic regression. Multivariate logistic regression stratified by sex was used to estimate the odds of ANA positivity and the 95% confidence interval (CI) for diabetes, controlling for age, race and BMI and multimorbidity in a second model controlling for age and race. All analyses were performed using SAS version 9.3 (SAS Institute, Inc., Cary, NC).

3. Results

Table 1 shows the frequency of ANA and age-adjusted associations of ANA with demographic characteristics. The sample was 51% women and 27% black. The overall frequency of ANA was higher in women (15.2%) than men (8.8%) (OR=1.81; 95% CI 1.14, 2.87). A higher percentage of black participants were ANA seropositive than non-black participants, but this difference was not statistically significant. The mean age at ANA assessment was 73 years. The age distribution of ANA prevalence varied significantly by sex (interaction p<0.05, Figure 1).

Table 1.

Frequency of ANA and age-adjusted associations of ANA with demographic characteristics in the Baltimore Longitudinal Study of Aging

Variable	Total Population				Women				Men
	N	NANA+ (%)	ORa	95% CI	N	N ANA+ (%)	ORb	95% CI	N	N ANA+ (%)	ORb	95% CI
Total	745	90(12.1)			382	58(15.2)			363	32 (8.8)
Sex
Male	363	32 (8.8)	Reference
Female	382	58 (15.2)	1.81	1.14,2.87
Race
Non-Black	540	61 (11.3)	Reference		271	40 (14.8)	Reference		269	21 (7.8)	Reference
Black	205	29 (14.2)	1.22	0.75, 2.00	111	18 (16.2)	1.01	0.54, 1.87	94	11 (11.7)	1.81	0.80, 4.09
Enrollment Age
Before 70	670	83 (12.4)	Reference		342	54 (15.8)	Reference		328	29 (8.8)	Reference
70+	75	7 (9.3)	0.92	0.33,2.55	40	4 (10)	0.94	0.24, 3.66	35	3 (8.6)	0.82	0.18,3.71
Education (N=746)
<High School	51	6 (11.8)	1.08	0.44, 2.68	28	5 (17.9)	1.32	0.46, 3.79	23	1 (4.4)	0.48	0.06,3.76
Some College	112	14 (12.5)	1.11	0.59, 2.09	66	10 (15.2)	1.09	0.50, 2.38	46	4 (8.7)	1.07	0.34,3.35
College Degree	155	21 (13.6)	1.19	0.69, 2.07	85	13 (15.3)	0.99	0.48,2.01	70	8 (11.4)	1.46	0.61,3.52
> College	427	49 (11.5)	Reference		203	30 (14.8)	Reference		224	19 (8.5)	Reference
Income (N = 573)
< $50,000	83	7 (8.4)	0.68	0.29, 1.55	42	3 (7.1)	0.45	0.13, 1.56	41	4 (9.8)	0.98	0.32,3.04
Exceeds $50,000	490	64 (13.1)	Reference		234	40 (17.1)	Reference		256	24 (9.4)	Reference
Income Meets Needs (N = 642)
Poorly or Fairly Well	181	23 (12.7)	1.08	0.64, 1.82	100	13 (13.0)	0.84	0.42, 1.69	81	10 (12.4)	1.69	0.75,3.82
Very Well	461	54 (11.7)	Reference		221	35 (15.8)	Reference		240	19 (7.9)	Reference
Lifetime Hormone Replacement Therapy (post-menopausal)
Current					24	5 (20.8)	1.98	0.61,6.40
Past					170	23 (13.5)	1.15	0.52, 2.54
Never					103	13 (12.6)	Reference

^aadjusted for sex, age spline and race

^badjusted for sex-specific age spline and race

95% CI: 95% confidence interval

Figure 1.

Percent ANA positive (3+ or 4+) at 1:160 dilution by age group and sex in the Baltimore Longitudinal Study of Aging, N = 746; ¹p-value for interaction <0.05.

Total population and sex-specific distributions of the medical conditions used to generate the multimorbidity variable by ANA status are reported in Supplemental Table 3. Results from total population and sex-stratified logistic regression models for the association between ANA positivity and diabetes and multimorbidity are shown in Table 2.

Table 2.

Association (odds ratios and 95 % confidence intervals) between ANA+ and Diabetes and Multimorbidity in the Baltimore Longitudinal Study of Aging

Diabetes
	Total N (%)	ANA+ N (%)	OR ANA+3		95% CI
Total Study1	192(26.6)	23 (12.0)	1.30	0.76	2.22
Women2	71 (19.3)	16 (22.5)	2.06	1.04	4.07
Men2	131 (34.3)	7 (5.8)	0.60	0.24	1.48
Multimorbidity
	Total N (%)	ANA+ N (%)	OR ANA+		95% CI
Total Study1	166 (23.4)	22(13.3)	1.58	0.88	2.85
Women2	46 (12.7)	11(23.9)	2.47	1.11	5.50
Men2	120 (34.7)	11(9.2)	1.06	0.46	2.46

Total N (%) is the number and percent of cases with diabetes or multimorbidity; ANA+ N (%) is the number and percent of participants with diabetes or multimorbidity who had ANA antibodies.

¹adjusted for sex, age spline and race

²adjusted for sex-specific age spline and race

³additionally adjusted for BMI

Among ANA positive women 22.5% had diabetes and among ANA positive men 5.8% had diabetes. Sex-specific analyses revealed a positive association between ANA positivity and diabetes in women and a negative association in men. Women with diabetes had higher odds of ANA positivity than women without diabetes (OR = 2.06, 95% CI: 1.04, 4.07) whereas in men, no association was found (OR = 0.60, 95% CI: 0.24, 1.48) (p-interaction = 0.01). Men were more likely to report 2 or more health conditions (35%) than women (13%), however, 24% of ANA+ women had multimorbidity compared to only 9.2% of ANA positive men. Similar to the diabetes finding, no association between ANA positivity and multimorbidity was observed in the total population. Among women, those with a multimorbidity score of 2 or greater had 2.47 times the odds of ANA (95% CI: 1.11, 5.50) compared to women with 1 or 0 conditions.

4. Discussion

Our study makes an important contribution to understanding the intersection of sex-specific aging, ANA and morbidity. Some studies have reported a linear, dose-response relationship between age and ANA over the lifecourse,^5,7 and others have reported no association between age and ANA in older individuals.¹⁸ However, understanding patterns of ANA prevalence by sex at older ages is limited by a lack of research in healthy older adults and the absence of reporting of ANA prevalence by age and sex in the few studies that have been conducted. In this study of middle-age and older individuals free of autoimmune disease, we observed a non-linear relationship between age and ANA that varied by sex. Clear sex differences in ANA seropositivity at younger ages (48–59) appeared to converge in the oldest individuals (80+). These results, consistent with other findings of immune response differences by sex⁸, suggest that future studies in middle and older aged adults should explore the possibility of effect modification of the age-ANA relationship by sex.

Observations of increased ANA with advancing age from previous studies suggest immunosenescence and increased apoptosis as drivers of these findings in the absence of autoimmune disease.¹⁹ Aging-associated immune dysfunction and imbalance, due to remodeling or dysregulation of the immune system¹¹, including B-cell hyperactivity and impaired regulatory activities of the adaptive immune system,^20,21 are reflected by well-documented patterns of decreased functional immunity and increased chronic inflammation.²² Increased aging-associated ANA may also arise in response to cellular damage in other disease processes.^11,19,21 In community-based and clinical samples, ANA has been associated with several non-autoimmune chronic conditions, including cardiovascular disease, renal insufficiency and cancer.²³ As only a few studies have been conducted in well-characterized aging individuals, the relationship between other non-autoimmune morbidities and ANA positivity remains unclear.⁵ Elevated inflammation is thought to be a driver of many chronic diseases.^14,24,25 The presence of autoimmune markers, such ANA and other autoantibodies, may reflect pathways by which chronic disease generate immune dysregulation or a pathophysiologic pathway by which elevated inflammation contributes to chronic morbidity.^10,20

Type 2 diabetes (T2D) is commonly thought of as a metabolic disorder, but recent studies suggest that T2D is heterogeneous and autoimmunity and insulin resistance combined may contribute to disease development.¹² Indeed, therapies targeting inflammation and immunomodulation have improved T2D metabolic profiles¹², possibly by preventing autoimmune pathways and B-cell function decline.¹² Further, chronic inflammation is a common feature of T2D, which may promote autoimmunity through cellular damage, tissue damage and defective apoptosis.¹² The sex differences observed in the association between ANA and diabetes in this study are consistent with research showing that women are more likely to have an inflammatory diabetes phenotype compared to men.^8,9 Thus, our findings are consistent with the possibility that T2D phenotype observed in women may be more likely to have an autoimmunity component than men, resulting in the positive association between ANA and diabetes among women. Although Type 1 diabetes is the most commonly recognized autoimmune phenotype, at enrollment BLSA study participants were older and free of chronic diseases including type 1 diabetes. We lack information on other autoantibodies that would be needed to identify cases of latent autoimmune diabetes.^26,27

In this study, men were more likely than women to report two or more health conditions; however, multimorbidity was associated with ANA seropositivity in women only. Women may be more likely than men to have immune dysfunction leading to the appearance of ANA given sex differences in pro-inflammatory profiles and may also experience increased cell death due to unchecked ROS-induced DNA damage as a result of estrogen loss during menopause.⁹ Women in our study are overwhelmingly post-menopausal and only five individuals were currently on hormone replacement therapy. Future research on differences in age-associated changes in immune function by sex in the context of chronic low-grade inflammatory disease are needed to clarify complex disease processes resulting in ANA production.

This study has limitations. Only a subset of BLSA participants were tested for ANA positivity, limiting our analysis to investigating cross-sectional associations. Determining whether T2D and multimorbidity in this study contribute to ANA seropositivity or result from it will be a crucial next step. Further study is also warranted for other possible cellular damage mechanisms, such as infections/vaccinations received over the life course, for which we did not have data. The BLSA is a study of normative aging and, due to study requirements, participants are generally healthier than others their age. This selection criterion has the potential to bias our results as individuals who enter the study at an older age are healthier on average than individuals who entered at a younger age. Although those with enrolled at age 70 or older tended to have a slightly lower frequency of ANA seropositivity (9.2%) it did not differ from the rate observed in those who enrolled before age 70 (12.4%). Additionally, adjusting for BLSA enrollment before age 70 did not meaningfully alter results.

Despite these limitations, this study addresses an important gap in the literature by examining sex-specific differences in the relationship between age and ANA as well as identifying sex-specific associations between ANA and diabetes and multimorbidity in a relatively large aging cohort of autoimmune-disease free individuals. Our primary analyses used a high dilution and a stringent cut-point to define ANA seropositivity, reducing the likelihood of false positives and lending strength to our results. While absolute ANA prevalence cannot be compared with other studies due to inter-laboratory assay variability^11,28,29 and non-uniform dilution levels for analysis, examining ANA prevalence relative to age and morbidity should be internally valid. In comparison to two other studies focused on ANA in older samples^11,30, our study is larger and has a broader age distribution as well as an internal referent group for examining age trends in ANA. This analysis was primarily exploratory and our results provide guidance for future work on sex-specific associations between ANA and age-related morbidities with inflammatory etiology.

Higher levels of ANA in autoimmune-disease free individuals may be a result of chronic inflammation and impaired regulatory processes, which normally promotes tolerance to self-antigens. The prevalence of ANA in healthy middle aged and older individuals is likely not a precursor to autoimmune disease as late-onset of autoimmune diseases such as SLE is uncommon^11,31, but instead may be a marker of autoimmune processes contributing to the pathogenesis of age-associated diseases. We report several novel findings suggesting further research on sex-specific associations between diabetes, multimorbidity and ANA is needed. Given the importance of maintaining immune function in advancing age, understanding the relationship of ANA with age, sex, aging-related morbidities, functional status and frailty may lend insight into healthy aging.

Highlights.

• Sex differences in ANA seropositivity at younger ages appeared to converge in the oldest individuals

• Women with multimorbidity have higher odds of ANA

• Understanding the contribution of ANA to morbidity may lend insight into healthy aging

Appendix A

Disease-modifying anthirheumatic drugs (DMARDs):

Abatacept

Adalimumab

Anakinra

Azathioprine

Belimumab

Certolizumab pegol

Chloroquine

D-penicillamine

Gold compounds

Golimumab

Infliximab

Leflunomide

Quinacrine

Rituximab

Sulfasalazine

Tocilizumab