Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Dec 1.
Published in final edited form as: Semin Arthritis Rheum. 2017 May 25;47(3):376–383. doi: 10.1016/j.semarthrit.2017.05.011

Obesity and the Risk of Systemic Lupus Erythematosus among Women in the Nurses’ Health Studies

Sara K Tedeschi a, Medha Barbhaiya a, Susan Malspeis a, Bing Lu a, Jeffrey A Sparks a, Elizabeth W Karlson a, Walter Willett b,c, Karen H Costenbader a
PMCID: PMC5675759  NIHMSID: NIHMS879968  PMID: 28688713

Abstract

Background

Obesity is increasingly prevalent and related to increased risk of several autoimmune diseases, likely via generation of inflammatory adipokines. Prior studies have not evaluated obesity in relation to systemic lupus erythematosus (SLE) risk. We prospectively evaluated whether obesity was associated with increased SLE risk among women in the U.S. Nurses’ Health Study cohorts.

Methods

We conducted a prospective cohort study among 238,130 women in the Nurses’ Health Studies (NHS, 1976–2012; NHSII, 1989–2013). Incident SLE was confirmed by American College of Rheumatology 1997 criteria and validated through medical record review. Body mass index (BMI, kg/m2) was calculated at baseline and on biennial questionnaires. Cox proportional hazards models estimated HRs (95% CIs) for SLE by cumulative average BMI category [18.5 to <25 (normal [reference]), 25 to <30 (overweight), ≥30 (obese)], adjusting for potential time-varying confounders. Models were performed separately in each cohort; results were meta-analyzed. Sensitivity analyses used simple time-varying BMI, a four-year lag between exposure and SLE risk window to address potential reverse causation, and evaluated BMI at age 18 and weight change since age 18. A secondary analysis started follow-up in both cohorts at similar calendar years when the prevalence of obesity in the U.S. increased most dramatically [1988 (NHS)/1989 (NHSII)].

Results

We identified 153 NHS incident SLE cases and 115 incident NHSII cases during 5,602,653 person-years of follow-up. At baseline, 8.4% of women in NHS and 11.8% in NHSII were obese. Mean age at enrollment was 42.5 (SD 7.2) years in NHS and 34.4 (SD 4.7) years in NHSII. Cumulative average obesity was significantly associated with SLE risk in NHSII (HR 1.85 [1.17–2.91]), but not in NHS (HR 1.11 [0.65–1.87]) compared to normal BMI. In the meta-analysis of both cohorts, obesity was not significantly associated with increased risk of SLE (HR 1.46 [0.88–2.40]). Simple time-varying BMI and lagging the exposure window by four years produced similar findings to the primary analysis. In NHSII, a 10 pound gain between age 18 and enrollment slightly increased SLE risk (HR 1.09 [1.02–1.18]). In the secondary analysis starting follow-up of both cohorts at similar calendar year, the point estimate for obesity in NHS was higher than the primary analysis (HR 1.67 [0.81–3.45]).

Conclusion

We observed an 85% significantly increased risk of SLE among obese compared to normal BMI women in the more recent NHSII cohort, but no association was observed in the earlier NHS cohort. Secular trends in obesity may account for the differences between the two birth cohorts.

Keywords: systemic lupus erythematosus, body mass index, obesity

INTRODUCTION

Systemic lupus erythematosus (SLE) is a heterogeneous, chronic autoimmune disease that occurs predominantly in women. SLE is thought to develop in genetically-susceptible individuals after exposure to environmental risk factors. Specific or cumulative environmental exposures may contribute to the initial break in immune tolerance resulting in autoantibody formation and to the subsequent progression to clinical manifestations over time.1 In a prospective study of U.S. Armed Forces members who were asymptomatic at the time of blood draw, the mean interval between the detection of autoantibodies and onset of SLE was 3.3 years, and was as long as 9.4 years.2 Thus, the window during which environmental exposures may contribute to the development of SLE is wide.

Among women, a number of reproductive factors have been associated with increased SLE risk: oral contraceptive use,3,4 age ≤10 at menarche;3 and use of postmenopausal hormone replacement therapy.3 Parity and breastfeeding, however, have not been associated with SLE risk.3 Other environmental risk factors associated with increased SLE risk include current cigarette smoking5 and crystalline silica.6 Moderate alcohol intake (>0.5 drinks/day) has been associated with a decreased risk of incident SLE.7

The prevalence of obesity in the U.S. has been increasing over the past four decades, particularly since the late 1980s.8 In 2014, over 40% of U.S. adult women fulfilled the World Health Organization definition of obesity (body mass index [BMI] ≥30 kg/m2).9 Obesity is a serious public health problem, associated with increased risk for many chronic diseases including hypertension,10 type 2 diabetes,11 osteoarthritis,12 coronary artery disease,13 and cancer.14 Adipose tissue secretes inflammatory cytokines, termed adipokines, the presence of which may contribute to the development of chronic diseases in individuals with additional genetic or environmental risk factors. Adipose tissue also increases circulating estrogens via aromatase enzymes, which convert androgens to estrogens.15 Prospective cohort studies have identified associations between obesity and increased risk of other autoimmune diseases, such as rheumatoid arthritis,16–18 psoriatic arthritis,19,20 and type 1 diabetes.21,22 However, to our knowledge, no previous study has investigated an association between obesity and risk of incident SLE.

We hypothesized that chronic obesity was associated with increased risk for incident SLE among women, and tested this hypothesis in two prospective cohorts of women: the U.S. Nurses’ Health Study and Nurses’ Health Study II.

MATERIALS AND METHODS

Study design and population

The Nurses’ Health Study (NHS) is a prospective cohort that enrolled 121,700 women living in 11 U.S. states at its start in 1976, while Nurses’ Health Study II (NHSII) is a prospective cohort that enrolled 116,430 women living in 14 states in 1989. The cohorts differ by age at enrollment, calendar years of birth, and duration of follow-up and thus comprise two distinct birth cohorts. NHS participants were enrolled at ages 30–55 in 1976, born from 1921–1946, and have been followed for 36 years with most recent follow-up at ages 66–91. NHSII participants were enrolled at ages 25–42 in 1989, born from 1947–1964, and have been followed for 24 years with most recent follow-up at ages 49–62. Thus NHS is an earlier birth cohort exposed to lifestyle and environmental factors dating back to 1921, and NHSII is a more recent birth cohort with exposures dating back to 1947. Participants have completed mailed questionnaires at baseline and every two years in follow-up regarding lifestyle factors, health behaviors, and the development of new diseases. Follow-up rates have been high and only 5% of person-time has been lost to follow-up.23 The National Death Index is searched every two years for non-responders, to validate mortality.

Identification of cases

In this analysis, NHS participants were followed from 1976 through May 31, 2012, and NHSII participants were followed from 1989 through May 31, 2013. Methods for SLE case identification and validation according to ACR 1997 Classification Criteria have previously been reported.3,24 Two rheumatologists independently reviewed medical records for women who self-reported SLE, to confirm whether four of the 11 ACR criteria were met. We excluded participants self-reporting an existing diagnosis of SLE at cohort entry, and censored those who self-reported SLE in follow-up without confirmation by medical record review.

Exposures and covariates

Body mass index (BMI)

BMI was calculated as weight in kilograms divided by squared height in meters (kg/m2) using self-reported height from the baseline questionnaire and self-reported weight updated on biennial questionnaires. In a previous validation study among 140 NHS participants, self-reported weight and measured weight had high accuracy (r=0.97).25 Missing BMI values were imputed by assigning the last BMI carried forward, for up to two consecutive questionnaire cycles. As we hypothesized that long-term obesity, rather than short-term obesity, would be related to SLE risk, cumulative average BMI was the primary exposure. Cumulative average BMI was calculated as the average BMI over all questionnaire cycles prior to the current questionnaire cycle.

BMI in the setting of pregnancy was handled differently, given that weight gain during pregnancy is not primarily due to excess adipose tissue. For questionnaire cycles in which women reported being pregnant, we carried forward BMI from the prior cycle before calculating cumulative average BMI, consistent with prior analyses of BMI and risk of rheumatoid arthritis.16

We categorized cumulative average BMI according to World Health Organization (WHO) categories: normal (18.5 kg/m2 to <25 kg/m2 [reference]), overweight (25 kg/m2 to <30 kg/m2), and obese (≥30 kg/m2).26 Women with BMI <18.5 kg/m2 at baseline were excluded due to insufficient numbers of underweight SLE cases, thus the model was unstable if an underweight category were included the main exposure (n=3268 women in NHS, of whom 1 developed SLE; n=3949 women in NHSII, of whom 2 developed SLE). We did not include underweight women in the reference group due to concerns about different effects of underweight and normal body mass index on SLE risk. In the sensitivity analyses testing the relationship of BMI at age 18 and SLE risk we were able to include an underweight category (BMI <18.5 kg/m2), as there were a sufficient number of underweight cases at age 18 to stabilize the model. In an analysis of weight change from age 18 to baseline, we included women with BMI <18.5 kg/m2 at age 18 or at baseline, as the primary exposure was continuous weight change. In secondary analyses, simple time-varying BMI updated at each questionnaire cycle in WHO categories was the primary exposure. We also assessed SLE risk associated with simple time-varying and cumulative average BMI as continuous measures.

Covariates

Race was dichotomized as White vs. non-White, given that ~90% of NHS and NHSII participants self-report White race. We adjusted for time-varying covariates, updated at each biennial questionnaire, which have previously been associated either with SLE risk, obesity, or both. Age was treated as a continuous variable. U.S. Census-tract household median income was employed as a marker of socioeconomic status (< vs.≥ sample median $60,000). Cigarette smoking was categorized as never, past, or current.5 Cumulative average alcohol intake was categorized as none, >0 to <5 grams/day, or ≥5 grams/day.7 Oral contraceptive use was dichotomized as ever vs. never, and age at menarche was dichotomized as ≤10 vs. >10 years old.3 Menopausal status and post-menopausal hormone (PMH) use were categorized in three groups: pre-menopausal, post-menopausal/never used PMH, and post-menopausal/ever used PMH. We tested whether physical activity or husband’s educational level were confounders of the relationship between BMI and SLE, and found that neither of these variables changed the effect estimate of BMI category by >10%; thus we did not include them in multivariate models.

Statistical methods

Characteristics of NHS and NHSII participants at cohort entry were analyzed according to baseline BMI category. Among incident SLE cases, features of SLE were examined separately in each cohort according to cumulative average BMI category at date of SLE diagnosis. Cox proportional hazards models estimated hazard ratios (HR) and 95% confidence intervals (CI) for SLE risk by cumulative average BMI category; models were performed separately in each cohort. The proportional hazards assumption was tested using the Wald Chi square test for cumulative average BMI over the duration of follow-up in NHS and NHSII, and was met in each cohort. Models were initially adjusted only for age and questionnaire cycle, followed by multivariable models adjusted for time-varying covariates. Meta-analysis of NHS and NHSII HRs was performed using a DerSimonian-Laird random effects model.27

In sensitivity analyses, Cox proportional hazards models were re-run using modified exposures and modified windows of follow-up time. We first assessed the effect of shorter-term BMI elevations on SLE risk in “simple time-varying BMI” models. We also evaluated for reverse causation, that is, SLE symptoms resulting in weight change prior to clinical diagnosis, in separate models lagging the exposure time window for >4 years before SLE diagnosis.

We conducted four secondary analyses. Given secular changes in diet and obesity in the U.S., with a sharp rise in obesity prevalence in the general U.S. population in the late 1980s,8 we restricted the calendar years of follow-up in NHS to more closely match the calendar years of follow-up in NHSII. In this model NHS follow-up began in 1988, to parallel the 1989 cohort start date of NHSII; follow-up ended on the same dates described in the primary analysis. Of note, women in NHS were ages 42–67 in 1988. Second, in order to test whether age at SLE diagnosis modified the effect of obesity on SLE risk, we performed stratified analyses in each cohort to assess risk of SLE onset at age <50 years vs. ≥50 years , as several studies have investigated differences in SLE manifestations using this age cutoff for early- versus late-onset SLE.28–30 Third, we performed an analysis using BMI at age 18 as the exposure, as early life obesity could be related to risk of autoimmune disease in later life.16,20 Last, we evaluted weight change between age 18 and cohort enrollment as a risk factor for SLE, as weight gain during young adulthood could be related to increased risk for later development of autoimmune disease. Weight change was calculated as change in pounds divided by 10, in order to present results in 10 pound increments for ease of interpretation. We excluded women reporting pregnancy at age 18 or at cohort enrollment from this secondary analysis, as it was not possible to impute weight from prior cycles. For all analyses, we tested for a linear trend in SLE risk with increasing BMI, using BMI as a continuous variable.

All aspects of this study were approved by the Partners’ HealthCare Institutional Review Board.

RESULTS

We identified 153 incident SLE cases among 114,527 women in NHS, and 115 SLE cases among 109,033 women in NHSII; greater than 90% of women in both cohorts were White. At enrollment, compared to the NHS cohort, NHSII cohort participants were younger, consumed less alcohol, smoked less, were more often pre-menopausal, and had higher oral contraceptive use (Table 1). Mean BMI was also higher in NHSII than NHS participants at cohort entry, with 11.8% obese in NHSII vs. 8.4% in NHS reflecting secular changes in obesity rates in the U.S. Within each cohort, obese women had lower income, lower mean alcohol intake, and menarche at age >10 less frequently than did women with normal BMI.

Table 1.

Age-standardized baseline characteristics of 114,527 women in the Nurses’ Health Study (NHS, enrolled 1976) and 109,033 women in the Nurses Health Study II (NHSII, enrolled 1989)

NHS NHSII
Characteristics Baseline BMI, kg/m2 Baseline BMI, kg/m2
18.5 to <25 25 to <30 ≥30 18.5 to <25 25 to <30 ≥30
Number (%) 81,201 (70.9) 23,691 (20.7) 9,635 (8.4) 75,435 (69.2) 20,746 (19.0) 12,852 (11.8)
Mean age, years (SD)* 41.9 (7.2) 44.1 (7.0) 44.0 (6.9) 34.1 (4.7) 35.0 (4.6) 35.4 (4.5)
White (%) 93.5 91.8 91.3 92.1 90.8 90.9
Census-tract annual household median income ≥$60,000 (%) 51.4 43.5 38.0 47.5 39.2 32.3
Smoking status (%)
 Never 41.8 46.8 50.1 65.0 64.4 65.6
 Past 23.6 23.6 25.3 21.9 21.5 20.9
 Current 34.6 29.6 24.6 13.1 14.1 13.5
Mean alcohol intake, g/day (SD) 7.1 (11.0) 5.0 (9.6) 3.1 (7.8) 3.3 (5.8) 2.6 (5.8) 1.9 (4.9)
Menarche age ≤10 (%) 4.8 8.4 11.7 6.0 10.8 16.0
OC use, ever (%) 48.6 43.6 42.3 84.3 83.2 80.1
Menopausal status and PMH use (%)
 Pre-menopausal 68.7 67.3 67.7 94.5 92.8 91.8
 Post-menopausal, never used PMH 16.0 18.1 19.5 2.7 3.6 3.5
 Post-menopausal, ever used PMH 14.3 13.6 11.9 2.7 3.5 4.5
Open in a new tab

OC: oral contraceptive PMH: post-menopausal hormone

*

Not age-adjusted

Abbreviations: g/day=grams/day

Among the 268 SLE cases in both cohorts, >95% had a positive ANA and >50% had a positive anti-dsDNA antibody at the time of diagnosis (Table 2). The presence of renal involvement at diagnosis as defined by ACR 1997 SLE Classification Criteria31 was similar between cohorts. Cytopenias were more frequent in NHSII, and arthritis was more frequent in NHS. There were no clear differences in these SLE manifestations across cumulative average BMI categories in the two cohorts.

Table 2.

Characteristics of 268 SLE cases at diagnosis by cumulative average BMI at diagnosis

NHS (n=153) NHSII (n=115)
Characteristics Cumulative Average BMI, kg/m2 Cumulative Average BMI, kg/m2
18.5 to <25 25 to <30 ≥30 18.5 to <25 25 to <30 ≥30
N (% of cases) 88 (57.5) 47 (30.7) 18 (11.8) 67 (58.3) 18 (15.6) 30 (26.1)
Age in years, mean (SD) 51.7 (8.4) 56.0 (9.3) 56.2 (10.1) 41.6 (7.4) 46.9 (9.2) 45.1 (7.5)
White (%) 90.9 91.5 83.3 95.5 77.8 96.7
ANA positive (%) 95.5 95.7 94.4 98.5 100.0 100.0
Anti-dsDNA positive (%) 31.8 42.6 50.0 47.8 77.8 40.0
Renal involvement (%) 18.2 21.3 16.7 13.4 16.7 10.0
Cytopenia (%) 56.8 51.1 33.3 67.2 66.7 60.0
Arthritis (%) 76.1 80.9 77.8 62.7 77.8 73.3
Open in a new tab

Cumulative average BMI and SLE risk (Table 3)

Table 3.

SLE risk by cumulative average BMI category among 114,527 women in the Nurses’ Health Study (NHS, 1976–2012) and 109,033 women in the Nurses’ Health Study II (NHSII, 1989–2013)

Cumulative Average BMI, kg/m2
18.5 to <25 25 to <30 ≥30
Normal Overweight Obese p trend+
NHS n=153 cases
 Mean BMI, kg/m2 (SD) 22.2 (1.6) 27.0 (1.4) 33.8 (3.6)
 Cases/person-years 88/1,989,754 47/897,130 18/380,805
 Adjusted HR (95% CI)* 1.00 1.28 (0.89–1.84) 1.11 (0.65–1.87) 0.49
NHSII n=115 cases
 Mean BMI, kg/m2 (SD) 22.1 (1.6) 27.1 (1.4) 35.2 (4.8)
 Cases/person-years 67/1,375,334 18/569,188 30/390,442
 Adjusted HR (95% CI)* 1.00 0.70 (0.41–1.18) 1.85 (1.17–2.91) 0.02
NHS+NHSII meta-analysis
 Cases/person-years 155/3,365,088 65/1,466,318 48/771,247
 Adjusted HR (95% CI)* 1.00 0.97 (0.54–1.76) 1.46 (0.88–2.40) 0.60
Open in a new tab
*

HRs adjusted for age in months, questionnaire cycle, pregnancy during the questionnaire cycle (yes/no), race (White/non-White), census-tract household median income (<$60,000 vs. ≥$60,000), smoking (never/past/current), alcohol intake (none, >0 to <5, ≥5 grams/day), oral contraceptive use (ever/never), age at menarche (≤10 vs. >10 years), menopausal status and post-menopause hormone (PMH) use (pre-menopausal, post-menopausal/never used PMH, post-menopausal/ever used PMH)

+

p value for cumulative average BMI (continuous kg/m2) in multivariable Cox models

Mean BMI was higher among obese NHSII participants (35.2 [4.8] kg/m2) compared to obese NHS participants (33.8 [3.6] kg/m2). Among women with normal BMI and overweight women, the mean BMI was similar in NHS and NHSII. In the primary analysis, NHS women who were overweight (HR 1.28 [0.89–1.84]) and obese (HR 1.11 [0.65–1.87]) did not have a significantly increased SLE risk compared to women with normal BMI (Table 3). In NHSII, obesity was associated with a significantly increased risk of SLE [HR 1.85 (1.17–2.92)], while overweight was not associated with increased SLE risk [HR 0.70 [0.41–1.18]). In NHSII, but not NHS, we observed increased SLE risk as continuous cumulative average BMI increased (p=0.02). Meta-analysis of both cohorts indicated a statistically non-significant increased SLE risk among obese women [HR 1.46 (0.88–2.41)], and SLE risk was not related to continuous BMI (p=0.60).

Simple time-varying BMI and SLE risk

When we used simple time-varying BMI as the exposure, our results were similar although possibly slightly attenuated compared to the models using cumulative average BMI (Table 4a). In NHS, neither overweight (HR 1.15 [0.80–1.67]) nor obesity (HR 1.00 [0.62–1.60]) were associated with elevated SLE risk. In NHSII, we observed a 77% increased risk of SLE among obese women (HR 1.77 [1.12–2.80]) and a non-significantly elevated SLE risk among overweight women (HR 1.11 [0.69–1.79]). As a continuous variable, simple time-varying BMI was associated with increased SLE risk in NHSII (p=0.03). In meta-analysis of NHS and NHSII employing simple time-varying BMI, we observed a statistically non-significant increased SLE risk among overweight and obese women (HR 1.33 [0.76–2.34]) and SLE risk was not related to continuous BMI (p=0.76).

Table 4a.

Sensitivity analysis: simple time-varying BMI SLE risk by simple time-varying BMI category among 114,527 women in the Nurses’ Health Study (NHS, 1976–2012) and 109,033 women in the Nurses’ Health Study II (NHSII, 1989–2013)

Simple Updated BMI, kg/m2
18.5 to <25 25 to <30 ≥30
Normal Overweight Obese p trend+
NHS n=153 cases
 Mean BMI, kg/m2 (SD) 22.3 (1.6) 27.2 (1.4) 34.3 (4.2)
 Cases/person-years 78/1,640,173 48/953,596 25/546,064
 Adjusted HR (95% CI)* 1.00 1.15 (0.80–1.67) 1.00 (0.62–1.60) 0.30
NHSII n=115 cases
 Mean BMI, kg/m2 (SD) 22.1 (1.6) 27.2 (1.4) 35.7 (5.3)
 Cases/person-years 52/1,166,432 27/610,195 34/515,742
 Adjusted HR (95% CI)* 1.00 1.11 (0.69–1.79) 1.77 (1.12–2.80) 0.03
NHS+NHSII meta-analysis
 Cases/person-years 130/2,806,605 75/1,563,791 59/1,061,806
 Adjusted HR (95% CI)* 1.00 1.14 (0.85–1.52) 1.33 (0.76–2.34) 0.76
Open in a new tab

4 cases out of 170,454 person-years are missing simple time-varying BMI and are accounted for by a missing indicator, but are not presented in the table

*

HRs adjusted for age in months, questionnaire cycle, race (White/non-White), census-tract household median income (<$60,000 vs. ≥$60,000), smoking (never/past/current), alcohol intake (none, >0 to <5, 5 grams/day), oral contraceptive use (ever/never), age at menarche (≤10 vs. >10 years), menopausal status and post-menopause hormone (PMH) use (pre-menopausal, post-menopausal/never used PMH, post-menopausal/ever used PMH)

+

p value for simple time-varying BMI (continuous kg/m2) in multivariate Cox models

Lagged analyses using cumulative average BMI

To assess for reverse causation (e.g. SLE symptoms affecting BMI immediately prior to clinical diagnosis), we performed models requiring a two-cycle (>4 year) lag between cumulative average BMI assessment and SLE diagnosis. In NHS, we observed a non-significantly increased SLE risk for overweight and obese women (Table 4b). In NHSII, obesity was again associated with significantly increased SLE risk (HR 1.95 [1.20–3.15]) compared to normal BMI, while there was no association between overweight and SLE. Meta-analysis of the cohorts revealed a statistically non-significant elevated SLE risk among obese women (HR 1.48 [0.84–2.60]).

Table 4b.

Sensitivity analysis: two-cycle lag between cumulative average BMI assessment and SLE diagnosis

Cumulative Average BMI, kg/m2
18.5 to <25 25 to <30 ≥30
Normal Overweight Obese p trend+
NHS n=143 cases
 Mean BMI, kg/m2 (SD) 22.2 (1.6) 27.0 (1.4) 33.8 (3.6)
 Cases/person-years 89/1,908,660 37/832,768 17/349,424
 Adjusted HR (95% CI)* 1.00 1.05 (0.71–1.55) 1.09 (0.64–1.87) 0.59
NHSII n=103 cases
 Mean BMI, kg/m2 (SD) 22.0 (1.6) 27.1 (1.4) 35.2 (4.8)
 Cases/person-years 59/1,286,026 17/512,544 27/347,356
 Adjusted HR (95% CI)* 1.00 0.76 (0.44–1.31) 1.95 (1.20–3.15) 0.01
NHS+NHSII meta-analysis
 Cases/person-years 148/3,194,686 54/1,345,312 44/696,780
 Adjusted HR (95% CI)* 1.00 0.94 (0.68–1.29) 1.48 (0.84–2.60) 0.53
Open in a new tab
*

HRs adjusted for age in months, questionnaire cycle, race (White/non-White), census-tract household median income (<$60,000 vs. ≥$60,000), smoking (never/past/current), alcohol intake (none, >0 to <5, ≥5 grams/day), oral contraceptive use (ever/never), age at menarche (≤10 vs. >10 years), menopausal status and post-menopause hormone (PMH) use (pre-menopausal, post-menopausal/never used PMH, post-menopausal/ever used PMH)

+

p value for cumulative average BMI (continuous kg/m2) in multivariable Cox models

Following both cohorts over similar calendar years

Because the prevalence of obesity in the U.S. steeply rose during the late 1980s, coinciding with the start of NHSII enrollment, we performed an analysis in which we delayed the start of NHS follow-up to more closely match the follow-up period for NHSII, using cumulative average BMI as the exposure (Table 4c). Thus, we restricted follow-up in NHS women to 1988–2012, while NHSII women were followed 1989–2013 as in the primary analysis. (In 1988, 13.9% of NHS participants were obese compared to 8.4% at cohort baseline in 1976.) Among NHS participants, we observed increased SLE risk among overweight (HR 1.77 [1.03–3.04]) and obese (HR 1.67 [0.81–3.45]) women. We did not observe a linear trend for increasing SLE risk associated with continuous cumulative average BMI (p=0.45). The analysis for NHSII was identical to the primary analysis. Meta-analysis showed a significant 80% increased SLE risk for obese women (HR 1.80 [1.22–2.65]) compared to normal BMI and an increased SLE risk associated with continuous cumulative average BMI (p=0.01).

Table 4c.

Sensitivity analysis: similar calendar-years of follow-up for NHS and NHSII SLE risk by cumulative average BMI category among 99,336 women in the Nurses’ Health Study (NHS) followed 1988–2012, and 109,033 women in the Nurses’ Health Study II (NHSII) followed 1989–2013

Cumulative Average BMI, kg/m2
18.5 to <25 25 to <30 ≥30
Normal Overweight Obese p trend+
NHS n=66 cases
 Mean BMI, kg/m2 (SD) 22.4 (1.5) 27.1 (1.4) 33.8 (3.6)
 Cases/person-years 29/1,100,360 26/593,310 11/255,032
 Adjusted HR (95% CI)* 1.00 1.77 (1.03–3.04) 1.67 (0.81–3.45) 0.45
NHSII n=115 cases
 Mean BMI, kg/m2 (SD) 22.1 (1.6) 27.1 (1.4) 35.2 (4.8)
 Cases/person-years 67/1,375,334 18/569,188 30/390,442
 Adjusted HR (95% CI)* 1.00 0.70 (0.41–1.18) 1.85 (1.17–2.92) 0.02
NHS+NHSII meta-analysis
 Cases/person-years 96/2,475,694 44/1,162,498 41/645,474
Adjusted HR (95% CI)* 1.00 1.11 (0.44–2.76) 1.80 (1.22–2.65) 0.01
Open in a new tab
*

HRs adjusted for age in months, questionnaire cycle, race (White/non-White), census-tract household median income (<$60,000 vs. ≥$60,000), smoking (never/past/current), alcohol intake (none, >0 to <5, ≥5 grams/day), oral contraceptive use (ever/never), age at menarche (≤10 vs. >10 years), menopausal status and post-menopause hormone (PMH) use (pre-menopausal, post-menopausal/never used PMH, post-menopausal/ever used PMH)

+

p value for cumulative average BMI (continuous kg/m2) in multivariable Cox models

Age-stratified analyses

Among 162 women in NHS and NHSII with incident SLE at age <50 years, the HR for cumulative average obesity was 1.07 (0.41–2.80) in a meta–analysis. There were 106 women aged ≥50 at SLE diagnosis in NHS and NHSII; cumulative average obesity was associated with a non-significant increased risk of late-onset SLE (meta-analyzed HR 2.21 [0.87–5.67]). In meta-analysis, continuous cumulative average BMI did not increase SLE risk in either age stratum (p=0.97 for SLE diagnosis at age <50, p=0.29 for SLE diagnosis at age ≥50).

BMI at age 18

Among women with data available for BMI at age 18, there were a total of 256 SLE cases in NHS and NHSII. Of these, only four were obese; 38 were underweight (BMI <18.5 kg/m2), 200 had normal BMI, and 16 were overweight. Continuous BMI at age 18 was not associated with increased SLE risk (meta-analyzed p=0.47). Obesity at age 18 was not associated with elevated SLE risk (meta-analyzed HR 0.73 [0.27–1.99]).

Weight change from age 18 to cohort enrollment

This analysis included a total of 256 SLE cases in NHS and NHSII. In NHS, a gain of 10 pounds between age 18 and cohort entry was not associated with increased SLE risk (HR 1.01 [0.94–1.10]). However, in NHSII, a 10 pound gain between age 18 and enrollment slightly increased SLE risk (HR 1.09 [1.02–1.18]). Notably, among women who gained ≥20 pounds between age 18 and cohort entry, the amount of weight gain was higher in NHSII compared to NHS (mean 39.2 [SD 21.4] pound gain in NHSII vs. mean 35.9 [SD 17.0] pound gain in NHS). Meta-analysis of NHS and NHSII did not reveal increased SLE risk associated with a 10 pound weight gain (meta-analyzed HR 1.06 [0.98–1.14]).

DISCUSSION

In this prospective analysis of two very large cohorts of female nurses with detailed covariate data, obese women in the more recent NHSII cohort had an 85% increased risk for developing SLE compared to women with normal BMI. By contrast, we detected no significant increased SLE risk among obese women in the earlier NHS cohort in the primary analysis. In NHSII, both cumulative average obesity and simple time-varying obesity were associated with increased SLE risk, though the point estimate for obesity was higher in the cumulative average BMI analysis. This is the first prospective cohort study, to our knowledge, to investigate the relationship between obesity and SLE risk. We were able to assess this relationship in two large, nationwide cohorts of female nurses with detailed, updated covariate data from biennial questionnaires and over 5.6 million person-years of follow-up.

Elevated cumulative average BMI is a long-term measure of adiposity, associated with chronic exposure to inflammatory adipokines and circulating estrogens produced by adipose tissue.32,33 Given the importance of systemic inflammation and reproductive factors in the pathogenesis of SLE, there is therefore strong biologic rationale for obesity increasing SLE risk. Obese individuals have higher levels of C-reactive protein and soluble TNF-alpha receptor 2 than non-obese individuals.33 Obesity has been identified as a risk factor for several other autoimmune diseases, including rheumatoid arthritis, psoriasis, psoriatic arthritis, inflammatory bowel disease, and sarcoidosis.16,17,34–39 Murine models of SLE-prone mice (NZB/WF1 females) also support the hypothesis that obesity may be associated with increased SLE risk, as these mice have higher body weight and greater visceral fat compared to control mice.40

Given that SLE pathogenesis is likely to develop slowly over years due to interactions between environmental exposures and genetic factors that lead to autoimmunity, we hypothesized that long-term exposure to excessive adipose tissue would be associated with increased SLE risk. After controlling for potential hormonal confounders including menopausal status, post-menopausal hormone use, and oral contraceptive use, both cumulative average obesity and simple time-varying obesity remained significant risk factors for SLE in the more recent NHSII cohort. In lagged analyses, we also detected a significant relationship between cumulative average obesity and SLE risk in the NHSII cohort, suggesting that this association was not an artifact of reverse causation. Since we found similar results in NHSII through a variety of different methods, the positive results are less likely related to chance.

There are a number of possible explanations for the disparate results in NHS and NHSII. Compared to NHSII women, NHS women enrolled at an earlier calendar year, participants were older at enrollment and born in the 1920s-1940s, were followed throughout middle and older ages up to age 91, obesity prevalence was lower throughout their most susceptible years, and mean BMI was lower among obese women. The peak incidence of SLE is among women of reproductive age;41 thus one possibility is that we observed a greater number of younger-onset SLE cases in NHSII due to cohort demographics at enrollment, and that age at diagnosis modifies the effect of obesity on SLE risk. Neither obesity at age 18 nor BMI at age 18 as a continuous measure was associated with increased SLE risk; however, this secondary analysis had limited power to detect an association due to the very small number of obese women at age 18 who later developed SLE after NHS or NHSII enrollment. Additionally, in a stratified analysis (age <50 vs. ≥50 years at SLE diagnosis), obesity did not significantly increase SLE risk among younger women.

Secular trends in obesity are a more likely explanation for the differing results in the NHS and NHSII cohorts. In 1976, when NHS women enrolled, U.S. obesity prevalence was lower, and mean BMI among obese women was lower, compared to 1989, the year of NHSII enrollment.8 Therefore, we may have lacked power to detect a relationship between obesity and SLE in NHS due to the lower prevalence of obesity among cases. In a secondary analysis, the follow-up periods were more closely matched (1988–2012 for NHS, 1989–2013 for NHSII). During these years of follow-up, obesity prevalence in NHS was more closely matched to obesity prevalence in NHSII. Additionally, the point estimate for obesity in NHS (HR 1.67 [0.81–3.45]) was closer in magnitude to the NHSII point estimate (HR 1.85 [1.17–2.92]), and after meta-analysis, obesity was significantly associated with risk of SLE (HR 1.80 [1.22–2.65]) compared to normal BMI. In our secondary analysis examining weight change between age 18 and cohort enrollment as a risk factor for SLE, we observed a significant risk in NHSII but not in NHS. This may be due to the greater weight gain among the NHSII participants, consistent with secular trends in obesity during the calendar years of NHSII enrollment.

Our study has some limitations, including that we cannot infer causation in this observational study. We do not have information on incident SLE at age <30 in NHS or age <25 in NHSII, to study the role of BMI in early-onset SLE. Our definition of SLE was stringent and required confirmation of at least four of 11 ACR 1997 SLE Classification Criteria. We were also not able to control for all potential confounders of the relationship between obesity and SLE. However, we were able to control for known risk factors for SLE and other potential confounders such as age, race, smoking, alcohol use, income as a marker of socioeconomic status, and reproductive and hormonal factors. We measured obesity using BMI rather than body composition measures, the latter of which may be more sensitive for capturing the risks of obesity in SLE patients.42 However, our study focused on SLE risk among women without SLE at the time of exposure assessment. We included U.S. Census-tract household median income as an indicator of socioeconomic status, though other socioeconomic factors may be important in the pathogenesis of SLE. Our analysis included female nurses who were older, employed and mostly White, and therefore our results may not apply to more diverse populations.

CONCLUSION

We identified an 85% increased risk of SLE among obese women in NHSII, which started enrollment in 1989, and a non-significant increase in SLE risk among obese women in NHS, which enrolled women in 1976. Overweight (BMI 25 to <30 kg/m2) compared to normal BMI (18.5 to <25 kg/m2) was not associated with increased SLE risk in either cohort, although significant linear relationships with increasing BMI were observed in NHSII. When the NHS cohort was delayed to start follow-up in the late 1980s, when U.S. obesity prevalence started increasing dramatically, meta-analysis of the two cohorts suggested an 80% increased risk of SLE among obese women. Further investigation into secular trends in both obesity and dietary factors in relation to SLE risk observed in recent years is warranted.

Acknowledgments

The authors would like to thank the participants of the Nurses’ Health Studies; the Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School; and the funding sources supporting this work: NIH K24 AR066109, R01 AR057327, L30 AR070514, L30 AR066953, K23 AR069688. The NHS was supported by National Institutes of Health (NIH) grants AR049880, AR052403, AR059073, AR066109, CA186107, and CA176726. Dr. Barbhaiya and Dr. Sparks receive support from the Rheumatology Research Foundation Scientist Development Awards. Dr. Tedeschi receives support through the Lupus Foundation of America Career Development Award. The funders had no rule in study design, data collection, analysis, decision to publish, or preparation of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the view of the National Institutes of Health.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • 1.Olsen NJ, Karp DR. Autoantibodies and SLE: the threshold for disease. Nature Reviews Rheumatology. 2014;10:181–6. doi: 10.1038/nrrheum.2013.184. [DOI] [PubMed] [Google Scholar]
  • 2.Arbuckle MR, McClain MT, Rubertone MV, et al. Development of autoantibodies before the clinical onset of systemic lupus erythematosus. The New England Journal of Medicine. 2003;349:1526–33. doi: 10.1056/NEJMoa021933. [DOI] [PubMed] [Google Scholar]
  • 3.Costenbader KH, Feskanich D, Stampfer MJ, Karlson EW. Reproductive and menopausal factors and risk of systemic lupus erythematosus in women. Arthritis and Rheumatism. 2007;56:1251–62. doi: 10.1002/art.22510. [DOI] [PubMed] [Google Scholar]
  • 4.Bernier MO, Mikaeloff Y, Hudson M, Suissa S. Combined oral contraceptive use and the risk of systemic lupus erythematosus. Arthritis and Rheumatism. 2009;61:476–81. doi: 10.1002/art.24398. [DOI] [PubMed] [Google Scholar]
  • 5.Costenbader KH, Kim DJ, Peerzada J, et al. Cigarette smoking and the risk of systemic lupus erythematosus: a meta–analysis. Arthritis and Rheumatism. 2004;50:849–57. doi: 10.1002/art.20049. [DOI] [PubMed] [Google Scholar]
  • 6.Finckh A, Cooper GS, Chibnik LB, et al. Occupational silica and solvent exposures and risk of systemic lupus erythematosus in urban women. Arthritis and Rheumatism. 2006;54:3648–54. doi: 10.1002/art.22210. [DOI] [PubMed] [Google Scholar]
  • 7.Barbhaiya M, Lu B, Sparks JA, et al. Influence of Alcohol Consumption on the Risk of Systemic Lupus Erythematosus among Women in the Nurses' Health Study Cohorts. Arthritis Care & Research. 2016 doi: 10.1002/acr.22945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Bray GA, Nielsen SJ, Popkin BM. Consumption of high-fructose corn syrup in beverages may play a role in the epidemic of obesity. The American Journal of Clinical Nutrition. 2004;79:537–43. doi: 10.1093/ajcn/79.4.537. [DOI] [PubMed] [Google Scholar]
  • 9.Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in Obesity Among Adults in the United States, 2005 to 2014. JAMA. 2016;315:2284–91. doi: 10.1001/jama.2016.6458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Landsberg L, Aronne LJ, Beilin LJ, et al. Obesity-related hypertension: pathogenesis, cardiovascular risk, and treatment--a position paper of the The Obesity Society and The American Society of Hypertension. Obesity (Silver Spring) 2013;21:8–24. doi: 10.1002/oby.20181. [DOI] [PubMed] [Google Scholar]
  • 11.Hu Y, Bhupathiraju SN, de Koning L, Hu FB. Duration of obesity and overweight and risk of type 2 diabetes among US women. Obesity (Silver Spring) 2014;22:2267–73. doi: 10.1002/oby.20851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Thijssen E, van Caam A, van der Kraan PM. Obesity and osteoarthritis, more than just wear and tear: pivotal roles for inflamed adipose tissue and dyslipidaemia in obesity-induced osteoarthritis. Rheumatology. 2015;54:588–600. doi: 10.1093/rheumatology/keu464. [DOI] [PubMed] [Google Scholar]
  • 13.Jensen MK, Chiuve SE, Rimm EB, et al. Obesity, behavioral lifestyle factors, and risk of acute coronary events. Circulation. 2008;117:3062–9. doi: 10.1161/CIRCULATIONAHA.107.759951. [DOI] [PubMed] [Google Scholar]
  • 14.Vucenik I, Stains JP. Obesity and cancer risk: evidence, mechanisms, and recommendations. Ann N Y Acad Sci. 2012;1271:37–43. doi: 10.1111/j.1749-6632.2012.06750.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Nelson LR, Bulun SE. Estrogen production and action. Journal of the American Academy of Dermatology. 2001;45:S116–S24. doi: 10.1067/mjd.2001.117432. [DOI] [PubMed] [Google Scholar]
  • 16.Lu B, Hiraki LT, Sparks JA, et al. Being overweight or obese and risk of developing rheumatoid arthritis among women: a prospective cohort study. Annals of the Rheumatic Diseases. 2014 doi: 10.1136/annrheumdis-2014-205459. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Crowson CS, Matteson EL, Davis JM, 3rd, Gabriel SE. Contribution of obesity to the rise in incidence of rheumatoid arthritis. Arthritis Care & Research. 2013;65:71–7. doi: 10.1002/acr.21660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.de Hair MJ, Landewe RB, van de Sande MG, et al. Smoking and overweight determine the likelihood of developing rheumatoid arthritis. Annals of the Rheumatic Diseases. 2013;72:1654–8. doi: 10.1136/annrheumdis-2012-202254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Love TJ, Zhu Y, Zhang Y, et al. Obesity and the risk of psoriatic arthritis: a population-based study. Annals of the Rheumatic Diseases. 2012;71:1273–7. doi: 10.1136/annrheumdis-2012-201299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Li W, Han J, Qureshi AA. Obesity and risk of incident psoriatic arthritis in US women. Annals of the Rheumatic Diseases. 2012;71:1267–72. doi: 10.1136/annrheumdis-2011-201273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Harpsoe MC, Basit S, Andersson M, et al. Body mass index and risk of autoimmune diseases: a study within the Danish National Birth Cohort. International Journal of Epidemiology. 2014;43:843–55. doi: 10.1093/ije/dyu045. [DOI] [PubMed] [Google Scholar]
  • 22.Viner RM, Hindmarsh PC, Taylor B, Cole TJ. Childhood body mass index (BMI), breastfeeding and risk of Type 1 diabetes: findings from a longitudinal national birth cohort. Diabet Med. 2008;25:1056–61. doi: 10.1111/j.1464-5491.2008.02525.x. [DOI] [PubMed] [Google Scholar]
  • 23.Chen WY, Rosner B, Hankinson SE, Colditz GA, Willett WC. Moderate alcohol consumption during adult life, drinking patterns, and breast cancer risk. JAMA. 2011;306:1884–90. doi: 10.1001/jama.2011.1590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Karlson EW, Sanchez-Guerrero J, Wright EA, et al. A connective tissue disease screening questionnaire for population studies. Ann Epidemiol. 1995;5:297–302. doi: 10.1016/1047-2797(94)00096-c. [DOI] [PubMed] [Google Scholar]
  • 25.Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Willett WC. Validity of self-reported waist and hip circumferences in men and women. Epidemiology. 1990;1:466–73. doi: 10.1097/00001648-199011000-00009. [DOI] [PubMed] [Google Scholar]
  • 26. [Accessed 20 Nov 2015];Overweight and obesity. at http://www.who.int/gho/ncd/risk_factors/overweight_text/en/
  • 27.DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986;7:177–88. doi: 10.1016/0197-2456(86)90046-2. [DOI] [PubMed] [Google Scholar]
  • 28.Merola JF, Bermas B, Lu B, et al. Clinical manifestations and survival among adults with (SLE) according to age at diagnosis. Lupus. 2014;23:778–84. doi: 10.1177/0961203314526291. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Lin H, Wei JC, Tan CY, et al. Survival analysis of late-onset systemic lupus erythematosus: a cohort study in China. Clinical Rheumatology. 2012;31:1683–9. doi: 10.1007/s10067-012-2073-6. [DOI] [PubMed] [Google Scholar]
  • 30.Alonso MD, Martinez-Vazquez F, Diaz de Teran T, et al. Late-onset systemic lupus erythematosus in Northwestern Spain: differences with early-onset systemic lupus erythematosus and literature review. Lupus. 2012;21:1135–48. doi: 10.1177/0961203312450087. [DOI] [PubMed] [Google Scholar]
  • 31.Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis and Rheumatism. 1997;40:1725. doi: 10.1002/art.1780400928. [DOI] [PubMed] [Google Scholar]
  • 32.Trayhurn P, Wood IS. Signalling role of adipose tissue: adipokines and inflammation in obesity. Biochem Soc Trans. 2005;33:1078–81. doi: 10.1042/BST0331078. [DOI] [PubMed] [Google Scholar]
  • 33.Panagiotakos DB, Pitsavos C, Yannakoulia M, Chrysohoou C, Stefanadis C. The implication of obesity and central fat on markers of chronic inflammation: The ATTICA study. Atherosclerosis. 2005;183:308–15. doi: 10.1016/j.atherosclerosis.2005.03.010. [DOI] [PubMed] [Google Scholar]
  • 34.Lahiri M, Luben RN, Morgan C, et al. Using lifestyle factors to identify individuals at higher risk of inflammatory polyarthritis (results from the European Prospective Investigation of Cancer-Norfolk and the Norfolk Arthritis Register--the EPIC-2-NOAR Study) Annals of the Rheumatic Diseases. 2014;73:219–26. doi: 10.1136/annrheumdis-2012-202481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Wesley A, Bengtsson C, Elkan AC, et al. Association between body mass index and anti-citrullinated protein antibody-positive and anti-citrullinated protein antibody-negative rheumatoid arthritis: results from a population-based case-control study. Arthritis Care & Research. 2013;65:107–12. doi: 10.1002/acr.21749. [DOI] [PubMed] [Google Scholar]
  • 36.Kumar S, Han J, Li T, Qureshi AA. Obesity, waist circumference, weight change and the risk of psoriasis in US women. J Eur Acad Dermatol Venereol. 2013;27:1293–8. doi: 10.1111/jdv.12001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Khalili H, Ananthakrishnan AN, Konijeti GG, et al. Measures of obesity and risk of Crohn's disease and ulcerative colitis. Inflammatory Bowel Diseases. 2015;21:361–8. doi: 10.1097/MIB.0000000000000283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Cozier YC, Coogan PF, Govender P, Berman JS, Palmer JR, Rosenberg L. Obesity and weight gain in relation to incidence of sarcoidosis in US black women: Data from the Black Women's Health Study. Chest. 2014 doi: 10.1378/chest.14-1099. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Tedeschi SK, Cui J, Arkema EV, et al. Elevated BMI and antibodies to citrullinated proteins interact to increase rheumatoid arthritis risk and shorten time to diagnosis: A nested case-control study of women in the Nurses' Health Studies. Seminars in Arthritis and Rheumatism. 2016 doi: 10.1016/j.semarthrit.2016.09.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ryan MJ, McLemore GR, Jr, Hendrix ST. Insulin resistance and obesity in a mouse model of systemic lupus erythematosus. Hypertension. 2006;48:988–93. doi: 10.1161/01.HYP.0000243612.02929.df. [DOI] [PubMed] [Google Scholar]
  • 41.Tedeschi SK, Bermas B, Costenbader KH. Sexual disparities in the incidence and course of SLE and RA. Clinical Immunology. 2013;149:211–8. doi: 10.1016/j.clim.2013.03.003. [DOI] [PubMed] [Google Scholar]
  • 42.Katz P, Gregorich S, Yazdany J, et al. Obesity and its measurement in a community-based sample of women with systemic lupus erythematosus. Arthritis Care & Research. 2011;63:261–8. doi: 10.1002/acr.20343. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES