Less than Seven Hours of Sleep per Night is Associated with Transitioning to Systemic Lupus Erythematosus

Abstract

Background

The role of sleep in the etiology of systemic lupus erythematosus (SLE) has not been well studied. We examined whether sleep duration was associated with subsequent transitioning to SLE in individuals at risk for SLE.

Methods

436 relatives of SLE patients who did not have SLE themselves at baseline were evaluated again an average of 6.3 (± 3.9) years later. Fifty-six individuals transitioned to SLE (≥ 4 cumulative ACR criteria). Sleep duration, medication use, and medical history were assessed by questionnaire; ACR criteria were confirmed by medical record review. Vitamin D was measured by ELISA. Generalized estimating equations, accounting for correlation within families, assessed associations between baseline sleep and the outcome of transitioning to SLE.

Results

Reporting sleeping less than seven hours per night at baseline was more common in those who subsequently transitioned than those who did not transition to SLE (55% versus 32%, p=0.0005; OR: 2.8, 95% 1.6–4.9). Those who transitioned to SLE were more likely to sleep less than seven hours per night than those who did not transition to SLE adjusting for age, sex and race (OR: 2.8, 95% CI 1.6–5.1). This association remained after individual adjustment for conditions and early symptoms that could affect sleep, including prednisone use vitamin D deficiency, and number of ACR criteria (OR: 2.0, 95% CI 1.1–4.2).

Conclusion

Lack of sleep may be associated with transitioning to SLE, independent of early clinical manifestations of SLE that may influence sleep duration. Further evaluation of sleeping patterns and biomarkers in at-risk individuals is warranted.

Introduction

Systemic lupus erythematosus (SLE) is a complex autoimmune disease characterized by autoantibody production, acute and chronic inflammation of various tissues, and fluctuating manifestations such as fever, rash and nephritis. The etiology and pathogenesis of SLE remains largely unidentified, but is most likely a combination of genetic and environmental risk factors.

Both quality and quantity of sleep have been shown to have significant impacts on health, including increased risk of metabolic diseases, hypertension, and mortality^{1, 2}. Both acute and chronic sleep disturbances and deprivation have reported associations with increased inflammatory markers^{3, 4}. Approximately one-third of adults in the United States sleep less than seven hours a night⁵, a threshold below which health impacts are manifested^{6, 7}. Sleep disturbances are commonly reported in SLE patients^8–14. Poor sleep quality in SLE has been correlated with fatigue ^{11, 15, 16}, depression¹³, and anxiety¹³. Prednisone use, commonly prescribed in SLE for other rheumatic disease symptoms, may have deleterious effects on sleep^{17, 18}. In addition, we have previously shown that vitamin D deficiency is associated with transitioning to SLE¹⁹, and lower vitamin D levels may be related to decreased sleep duration and quality^20–22, including in women with SLE ²³.

While sleep disturbances have been associated with disease activity⁸ and quality of life measures²⁴ in patients with SLE, whether the duration of sleep is related to risk of developing SLE has not been clearly established. A small case-control study of Japanese patients found that while self-reporting “sufficient sleep” was protective for SLE, no association was found for those reporting >7 hours of sleep per night²⁵. Two recent Taiwan population studies indicated that individuals, particularly females, diagnosed with a sleep disorder had an increased risk of SLE than those without a sleep disorder^{26, 27}. To further explore the association between sleep quantity and transitioning to SLE, we studied a cohort of individuals who reported having a family member with SLE, putting them at increased genetic risk for SLE^{28, 29}, but who did not meet SLE disease classification³⁰ at their baseline visit. We examined whether self-reported sleep duration at baseline was prospectively associated with transitioning to SLE at follow-up, independent of self-reported associated manifestations of SLE such as depression, fatigue, anxiety, or prednisone use.

Materials and methods

Participant information

All protocols were approved by the Institutional Review Boards at the University of Colorado Denver, the Oklahoma Medical Research Foundation (OMRF) and the University of Oklahoma Health Sciences Center. Study participants were enrolled in the Lupus Family Registry and Repository (LFRR) ³¹ and provided informed consent prior to enrollment. Letters were sent to 3823 individuals with a family member with SLE, who did not meet ≥ 4 American College of Rheumatology (ACR)³⁰ criteria for SLE at their prior (baseline) visit. These individuals were invited to enroll in a follow-up study to gather information regarding interim development of symptoms consistent with or a diagnosis of SLE. Compared to those not enrolled, the 436 individuals who enrolled were younger (mean age at follow-up 53.5 years vs 59.1 years), more likely to be female (84% vs 69%), and more likely to be European American (73.9% vs 55.4%). Of the 436 individuals enrolled, 402 individuals were first degree relatives (FDRs) and 34 individuals were second degree or more distant relatives of an individual with SLE.

Identical questionnaires and laboratory tests were completed at the baseline and follow-up examinations (mean time between contact 6.3 ± 3.9 years). Participants completed the Connective Tissue Disease Screening Questionnaire (CSQ)³², The SLE portion of the CSQ (SLE-CSQ) was scored using an algorithm based on ACR classification criteria³². Twenty-eight questions were grouped into 12 categories. If < 3 SLE-CSQ categories were true, then a person was categorized as unlikely SLE. If 3 of the 12 categories were true, then a person was categorized as possible SLE. If 4 of the 12 categories were true, then a person was categorized as probable SLE. Detailed demographic, environmental, clinical, and therapeutic information was collected by validated questionnaire³¹. Body mass index (BMI) was calculated as reported weight (kg)/height (m²). Peripheral blood samples were obtained from all study participants and processed for serum, plasma and DNA. Antinuclear antibodies (ANA) were detected using a HEp-2 indirect immunofluorescent assay (positive titer ≥1:120) and anti-dsDNA was detected using Crithidia luciliae (positive titer ≥1:30) according to manufacturer instructions (INOVA Diagnostics, San Diego, California, USA) in the College of American Pathologists-certified OMRF Clinical Immunology Laboratory. Plasma levels of 25-hydroxyvitamin D 25[OH]D were determined in duplicate using a commercial enzyme immunoassay (Immunodiagnostic Systems, Inc., Scottsdale, AZ) according to manufacturer instructions. Vitamin D deficiency was defined as 25[OH]D <20 ng/ml. Medical records were obtained and reviewed by a rheumatology-trained physician, physician assistant, or nurse for ACR classification criteria for individuals who reported signs and symptoms of SLE. Individuals were classified as having transitioned to SLE by the time of follow-up if they met ≥ 4 cumulative ACR criteria, in which medical record verified ACR criteria could accumulate towards the total count of ACR, and did not need to be all present at the same time. Fifty-six individuals were identified who transitioned to SLE at follow-up.

Measurement of Sleep Duration, Fatigue, Anxiety and Medication Use

Sleep duration in hours was assessed by the question “How many hours do you sleep at night on average?” Based on the Institutes of Medicine recommendation of 7–8 hours of sleep a night ³³ and the literature threshold for health impacts^5–7, we categorized our sleep outcome as <7 hours of sleep a night or ≥ 7 hours of sleep a night. Chronic fatigue, depression and anxiety were determined by whether the individual answered ‘yes’ when asked “whether you have had the condition or if a doctor has told you that you have the condition.” Prednisone use and sleep medication use within the last year were assessed by self-reported medication history.

Statistical analyses

All analyses were performed in SAS 9.4 (Cary, NC). T-tests for continuous variables and chi-square tests for categorical variables were used to determine differences between individuals who transitioned to SLE (n=56) and individuals who did not transition to SLE (n=380). Associations between sleeping less than seven hours a night and sleeping medication use, prednisone use, reported depression, reported fatigue, reported anxiety, SLE-CSQ category and number of ACR criteria at baseline were assessed by chi-square tests. We did not see any associations between baseline sleep and specific ACR classification criteria (e.g. arthritis, mucocutaneous, serositis, renal disorder, etc.). Within this cohort, individuals within the same family could be enrolled; family size ranged from 1–6 family members, with 23% of our population with at least one other family member enrolled. We therefore accounted for the correlation among these family members using generalized estimating equations, when assessing associations between a categorical measure of sleeping <7 hours a night at baseline and transitioning to SLE. All multivariable models were adjusted for age, sex, and race. BMI at baseline did not add significantly to these models, and was not included as a covariate. In addition, to determine if the association between sleep duration and SLE was independent of early clinical manifestations of SLE, we additionally adjusted for other sleep associated variables at baseline, including depression, prednisone use, chronic fatigue, SLE-CSQ category, number of ACR criteria and vitamin D deficiency. Due to the co-linearity of several of these factors (prednisone use, depression and chronic fatigue; and number of ACR criteria and SLE-CSQ category), we initially adjusted for these variables one at a time. A model was also included adjusting for prednisone use, number of ACR criteria and vitamin D deficiency. Odds ratios and 95% confidence intervals were determined for all models.

Results

We compared individuals who transitioned to SLE at follow-up to those who did not transition with respect to demographics, number of ACR criteria, and measures of sleep duration (Table 1). Individuals who transitioned had a greater number of confirmed ACR criteria, were more likely to be classified as probable or possible SLE by SLE-CSQ, and were more likely to be ANA positive at baseline compared to those who did not transition. Reported sleep medication and prednisone use at baseline were higher in those who transitioned compared to those who did not transition. A higher proportion of relatives who transitioned reported depression, fatigue, and anxiety at baseline. Individuals who transitioned to SLE reported fewer hours of sleep a night at baseline compared those who did not transition to SLE, and a higher proportion of individuals who transitioned to SLE reported sleeping less than seven hours a night (Table 1).

Table 1.

Baseline Characteristics of the Study Participants by Transitioned Status.

Variable (n, %)	Transitioned to SLE N=56	Did Not Transition N=380	p-value
Age at baseline (years): mean (median) ± sd	47.4 (48.5) ± 12.1	47.2 (48.0) ± 15.8	0.93
Time between contact (years): mean (median) ± sd	5.9 (5.1) ± 3.5	6.3 (5.4) ± 3.9	0.50
Sex: Female	49 (87.5%)	316 (83.2%)	0.41
Race: European American	43 (76.8%)	279 (73.4%)	0.55
African American	9 (16.1%)	55 (14.5%)
Othera	4 (7.1%)	46 (12.1%)
BMI at baseline: mean (median) ± SD	28.5 (28.5) ± 7.4	28.1 (26.9) ± 6.8	0.71
ANA positive at baseline	43 (76.8%)	183 (48.2%)	<0.001
SLE-CSQ Categories Baselineb:			<0.0001
Unlikely SLE	5 (8.9%)	248 (65.3%)
Possible SLE	5 (8.9%)	39 (10.3%)
Probable SLE	46 (82.1%)	93 (24.4%)
Number of ACR Criteria Baseline:			<0.0001
0–1c	7 (12.5%)	291 (76.5%)
2	23 (41.1%)	74 (19.5%)
3	26 (46.4%)	15 (4.0%)
Hours of Sleep Baselined: mean (median) ± sd	6.5 (6.0) ± 1.7	6.9 (7.0) ± 1.4	0.03
Sleep <7 hours at Baselined: Yes	31 (55.4%)	121 (31.8%)	0.0005
Sleep Medication Use at Baseline: Yes	15 (26.8%)	36 (9.5%)	0.0002
Prednisone Use at Baseline: Yes	33 (58.9%)	32 (8.4%)	<0.0001
Reported Depression at Baseline: Yes	36 (67.9%)	134 (37.1%)	<0.0001
Chronic Fatigue at Baseline: Yes	46 (86.8%)	103 (27.4%)	<0.0001
Anxiety Baseline: Yes	25 (49.0%)	89 (24.7%)	0.0003
Vitamin D Deficient: Yes	26 (46.4%)	126 (33.2%)	0.05

Significant differences are noted in bold. ANA, anti-nuclear antibodies. CSQ, Connective Tissue Disease Screening Questionnaire. ACR, American College of Rheumatology.

^aOther includes Native American (n=8), Asian (n=1) and Hispanic (n=2).

^bSLE-CSQ categories are based on responses to the SLE portion of the CSQ questionnaire. Twenty-eight questions were grouped into 12 categories. If the person indicated that < 3 categories were true, then a person was classified as unlikely SLE. If three of the 12 categories were marked as true, then the person was classified as possible SLE. If the person indicated that at least four of the 12 categories were true, then the person was classified as probable SLE.

^cThe bottom two categories (0 and 1) for number of ACR criteria were combined as no individuals in the transitioned group had 0 ACR criteria at baseline.

^dSleep data are missing for 23 individuals-3 who transitioned and 20 who did not transition.

Because specific conditions and early pre-clinical manifestations of SLE may be associated with sleep duration and therefore may be potential confounders, we examined whether presence of these characteristics were associated with sleeping less than seven hours a night at baseline (Table 2). Prednisone use, depression, and chronic fatigue, were all associated with sleeping less than seven hours a night. In addition, those who were vitamin D deficient, those with possible or probable SLE by SLE-CSQ, and those with 2 or 3 confirmed ACR criteria reported sleeping less than seven hours a night.

Table 2.

Reported Conditions and Associated Features of SLE by Sleep Status of the Study Participant at Baseline.

Variable (n, %)	Sleep <7 Hours N=152	Sleep ≥ 7 Hours N=261	p-value
Sleep Medication Use at Baseline: Yes	20 (13.2%)	30 (11.5%)	0.62
Prednisone Use at Baseline: Yes	34 (22.4%)	31 (11.8%)	0.005
Reported Depression at Baseline: Yes	73 (48.0%)	95 (36.3%)	0.02
Chronic Fatigue at Baseline: Yes	69 (45.4%)	76 (29.1%)	0.0008
Anxiety at Baseline: Yes	49 (32.2%)	64 (24.5%)	0.07
Number of ACR Criteria Baseline:			0.002
0–1	89 (58.6%)	191 (73.2%)
2	40 (26.3%)	54 (20.7%)
3	23 (15.1%)	16 (6.1%)
SLE-CSQ Categories Baseline:			0.003
Unlikely SLE	74 (48.7%)	165 (63.2%)
Possible SLE	13 (8.5%)	27 (10.3%)
Probable SLE	65 (42.8%)	69 (26.4%)
Vitamin D Deficient: Yes	73 (48.0%)	75 (28.7%)	<0.0001

Significant differences are noted in bold. CSQ, Connective Tissue Disease Screening Questionnaire. ACR, American College of Rheumatology

Adjusting for age, gender, and race, those who slept less than 7 hours a night at baseline were more likely to transition to SLE than those who slept more (adjusted OR: 2.8, 95% CI 1. 6–5.1). In order to examine whether this association is confounded by reported conditions and early pre-clinical features of SLE at baseline, we individually adjusted for baseline prednisone use, self-reported depression, self-reported chronic fatigue, vitamin D deficiency, SLE-CSQ category and number of ACR criteria (Table 3). The association between < 7 hours of sleep and transitioning to SLE remained significant in all models. In addition, we ran a final model adjusting for baseline prednisone use, number of ACR criteria, vitamin D deficiency. Sleeping < 7 hours per night was significantly associated with transitioning to SLE in this model as well, indicating there is an independent association between sleep duration and transitioning to SLE (Table 3). Sleep data was missing for 3 who transitioned and 20 who did not transition to SLE. Sensitivity analyses assuming those with missing sleep data had either all <7 hours of sleep per night or >7 hours of sleep per night did not significantly affect the magnitude of the association between sleep and transitioning to SLE. The final model adjusting for prednisone use, number of ACR criteria and vitamin D deficiency was also run excluding the second degree and more distantly related family members from the analysis. The magnitude of the association between sleeping < 7 hours per night and transitioning to SLE did not change (OR=2.1, 95% CI 0.9–4.6).

Table 3.

Less Than Seven Hours of Sleep a Night is Associated with Transitioning to SLE Independent of Associated Features of SLE.

		Transitioned to SLE (Odds Ratio (95% CI)
Baseline Characteristics	Base Model	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Age	–	1.0 (0.98–1.01)	1.0 (0.96–1.01)	1.0 (0.98–1.01)	1.0 (0.98–1.03)	1.0 (0.98–1.02)	1.0 (0.98–1.02)	1.0 (0.98–1.02)	0.99 (0.97–1.02)
Sex	–	1.3 (0.6–3.0)	0.8 (0.3–1.0)	0.9 (0.4–2.3)	0.7 (0.3–1.8)	1.1 (0.4–3.1)	0.7 (0.3–1.7)	1.3 (0.6–2.9)	1.0 (0.4–2.8)
Race (Non EA)	–	0.9 (0.4–2.0)	1.1 (0.4–3.0)	0.7 (0.3–0.8)	1.1 (0.4–2.7)	1.1 (0.5–2.4)	1.4 (0.5–3.3)	0.7 (0.3–1.7)	0.9 (0.4–2.1)
Sleep <7 hours	2.8 (1.6–4.9)	2.9 (1.6–5.1)	2.1 (1.1–4.2)	2.5 (1.4–4.6)	2.1 (1.1–3.9)	2.3 (1.2–4.5)	2.0 (1.1–3.8)	2.7 (1.5–4.8)	2.0 (1.1–4.2)
Prednisone Use	–	–	15.5 (7.7–31.4)	–	–	–	–	–	7.9 (3.6–17.6)
Self-reported depression	–	–	–	3.6 (1.9–6.7)	–	–	–	–	–
Self-reported chronic fatigue	–	–	–	–	18.3 (7.7–43.1)	–	–	–	–
ACR criteria:	–	–	–	–	–		–	–
2						13.4 (5.3–34.3)			7.3 (2.9–18.3)
3						180.8 (27.8–1174.4)			52.7 (8.3–335.7)
SLE-CSQ Categories:	–	–	–	–	–	–		–	–
Possible SLE							5.3 (1.3–21.0)
Probable SLE							28.1 (1.8–441.0)
Vitamin D Deficient	–	–	–	–	–	–	–	2.6 (0.8–8.8)	1.5 (0.3–7.7)

CSQ, Connective Tissue Disease Screening Questionnaire. ACR, American College of Rheumatology.

Of note, depression, and chronic fatigue were all significantly associated with transitioning to SLE, after adjusting for sleep. Interestingly, these associations with transitioning to SLE remained significant after additional adjusting for number of ACR classification criteria (chronic fatigue adjusted OR: 11.8, 95% CI 4.9–28.4; depression adjusted OR: 2.9, 95% CI 1.5–5.9).

Discussion

We found that sleep duration was associated with transitioning to SLE within a unique cohort of relatives of SLE patients without classified SLE at the baseline visit. This association was independent of several important variables which could indicate accompanying signs and symptoms of SLE, including: depression and prednisone use, both of which are increased in SLE and known to have an effect on sleep quality and architecture; and chronic fatigue, one of the most common and debilitating complaints in SLE and related to sleep.. While we have shown that being vitamin D deficient is associated with transitioning to SLE¹⁹, the association between sleep and transitioning to SLE was independent of vitamin D status. In addition, the association between sleep and transitioning to SLE was independent of SLE-CSQ category and the number of ACR criteria. In a full model adjusting for prednisone use, number of ACR criteria and vitamin D deficiency, sleep remained significantly associated with transitioning to SLE. This could indicate that while increasing signs and symptoms contribute to sleep disturbances, the contribution of sleep duration to transitioning to SLE may also be related to the deleterious health effects of disturbed and shortened sleep.

Sleep disturbances have been reported in SLE patients^8–14; however, none of these studies have examined the effect sleep has on risk of disease. One small case-control study found that while reporting >7 hours of sleep a day was not associated with SLE, reporting the relatively subjective outcome of getting “sufficient sleep” prior to SLE classification was inversely associated with SLE risk²⁵. However, given that cases in that study had been treated for SLE for up to 10 years, the data regarding sleep habits prior to SLE diagnosis may have been inaccurate or biased. Two Taiwanese population studies indicated that that individuals, diagnosed with a sleep disorder had an increased risk of SLE than those without a sleep disorder^{26, 27}. However, these studies did not account for medication use nor clinical manifestations of SLE, and their use of an insurance database may not adequately reflect individuals who do not seek medical care for their sleep disturbances.

The link between sleep duration and SLE etiology is feasible. Two studies utilizing the spontaneous (NZB/NZW) F₁ model of SLE revealed associations between sleep disturbances and disease. Sleep deprived mice demonstrated accelerated onset of lupus³⁴ through increased autoantibody production. Relatives of SLE patients are enriched in autoantibodies compared to controls ^{35, 36}. Given the SLE-related autoantibodies progressively accumulate³⁷, less sleep could accelerate the production of autoantibodies. In addition, altered sleep architecture was present both during progression to this form of murine lupus and during periods of increased disease activity³⁸. Sleep loss may deregulate immune responses^39–42, leading to increased inflammatory cytokines and increased white blood cell and natural killer cell counts. It could also lead to a break in self-tolerance through impaired regulatory T cells ⁴³. This pathophysiology is plausible, as obstructive sleep apnea has been associated with increased risk of autoimmune disease, including SLE and RA⁴⁴, as well as with increased biomarkers of immune function⁴⁵. Indeed, in this cohort of family members at increased genetic risk of SLE, SCF and TGFβ were shown to be independent predictors of transitioning to SLE ⁴⁶.

Increasing evidence supports a reciprocal relationship between sleep and immunity ^{47, 48}. An activated immune system alters sleep duration and quality, and sleep disturbances affect immune function. We saw a persistent magnitude of association when adjusting for accompanying signs and symptoms of SLE. This may indicate that less sleep duration is not merely an early symptom of SLE but may be involved in the pathogenesis of the disease prior to and during early SLE.

We also found that the associations between depression, anxiety, and chronic fatigue and transitioning to SLE remained significant independent of sleep. While poor sleep quality and architecture in SLE patients has been correlated with fatigue ^{11, 15, 16}, depression¹³, and anxiety¹³, this is one of the first studies to indicate that these symptoms occur prior to SLE classification. Interestingly, these associations with transitioning to SLE remained significant after additional adjusting for number of ACR classification criteria. Further study may explain whether anxiety, depression and chronic fatigue may also confer some independent risk other than as a symptom of early clinical manifestations of SLE.

One strength of the study is that the data are prospective. Consistent criteria were used at both visits using a validated questionnaire and examining medical records, making comparisons valid across time. However, one limitation to our study is the use of a self-reported measure of sleep, which is reported to have low correlation (0.43–0.45) with objective sleep measures such as actiography^{49, 50}. Self-reported sleep has been shown to over-report amount of sleep, possibly due to night-to-night variability and report of time spent in bed rather than actual sleep^{49, 50}. One study also found that people who report fair or poor health had no significant correlation between measured and reported sleep and report shorter hours of sleep on average⁵⁰. Therefore, it is possible that those reporting < 7 hours of sleep per night in this cohort are those who are experiencing poorer health. However, we did find that the association between sleep and transitioning to SLE was independent of number of ACR criteria and prednisone use, indicating that this may not be true in this population. In addition, we relied on self-reported measures of depression, sleep medication use, prednisone use, chronic fatigue and anxiety. Each of these measures were assessed by a single question, and more in-depth examination of effect of the quality of these measures could not be conducted. Family members who enrolled in this follow-up study were more likely to have a higher number of ACR criteria than those who did not enroll; therefore, they may be further along in the disease process and experiencing loss of sleep due to early signs/symptoms of SLE. However, we did see an association after adjusting for a greater number of ACR criteria, which could indicate that sleep may be a risk factor for transitioning to SLE independent of increasing number of signs or symptoms of disease. Further examination of the association between real-time, real-world monitoring of sleep and SLE using more objective measurements is warranted.

In summary, this is one of the first studies to determine that sleeping less than seven hours per night may be prospectively associated with transitioning to SLE, independent of several factors that may indicate early pre-clinical signs and symptoms of SLE, including prednisone use, the number of confirmed ACR criteria, and vitamin D deficiency. While increasing signs and symptoms may contribute to sleep disturbances, there may also be a contribution of the deleterious health effects of disturbed sleep on transitioning to SLE.