Systemic lupus erythematosus in a multiethnic US cohort (LUMINA): XXI. Disease activity, damage...

ARTHRITIS & RHEUMATISMVol. 52, No. 6, June 2005, pp 1655–1664DOI 10.1002/art.21048© 2005, American College of Rheumatology

Systemic Lupus Erythematosus in aMultiethnic US Cohort (LUMINA)

XXI. Disease Activity, Damage Accrual, and Vascular Events inPre- and Postmenopausal Women

Monica Fernandez,1 Jaime Calvo-Alen,1 Graciela S. Alarcon,1 Jeffrey M. Roseman,1

Holly M. Bastian,1 Barri J. Fessler,1 Gerald McGwin, Jr.,1 Luis M. Vila,2 Martha L. Sanchez,1

and John D. Reveille,3 for the LUMINA Study Group

Objective. To determine the differences in clinicalmanifestations, disease activity, damage accrual, andmedication use in systemic lupus erythematosus (SLE)patients as a function of menopausal status at diseaseonset.

Methods. Women with SLE as per the criteria ofthe American College of Rheumatology, with diseaseduration of <5 years and of Hispanic (Texas and Puerto

Rico ancestries), African American, and Caucasian eth-nicity, from LUMINA (LUpus in MInorities, NAture ver-sus nurture), a multiethnic, longitudinal cohort, werestudied. Menopause at the time of disease onset wasdefined as self-report of climacteric symptoms, and/oramenorrhea lasting >6 months, and/or oophorectomy,and/or increased follicle-stimulating hormone values forreproductive-age women, and/or treatment with hor-mone replacement therapy. Patients were divided intopremenopausal and postmenopausal categories. Socio-economic status, cumulative clinical manifestations,disease activity (at study entry or time 0, last visit, andover time), as measured by the Systemic Lupus ActivityMeasure, and damage accrual, as measured by theSystemic Lupus International Collaborating Clinics/American College of Rheumatology damage index (attime 0 and at last visit) were compared between the 2groups of women. Multivariable models were then ex-amined making adjustments for all possible known con-founders. Dependent variables in the models were renalinvolvement, damage accrual, arterial vascular events,and venous thrombosis.

Results. Five hundred eighteen women from theLUMINA cohort were included; 436 (84.2%) were pre-menopausal and 82 (15.8%) were postmenopausal. Dis-ease onset after menopause was more common amongCaucasians. Renal involvement was more common inpremenopausal women, whereas vascular arterialevents were more frequent in postmenopausal women.All other disease manifestations, as well as diseaseactivity, were comparable between both groups. Thepresence of damage accrual at time 0 and study end was

Supported by the NIH (grant R01-AR-42503 from the Na-tional Institute of Arthritis and Musculoskeletal and Skin Diseases andaward 1P20RR-11126 to the University of Puerto Rico from theNational Center for Research Resources), the General Clinical Re-search Center (grants M01-RR-02558 to the University of Texas andM01-RR-00032 to the University of Alabama, Birmingham [UAB]),The Mary Kirkland Scholars Award Program to UAB, and a PANLARFellowship Program to UAB.

1Monica Fernandez, MD, Jaime Calvo-Alen, MD, ScD, Gra-ciela S. Alarcon, MD, MPH, Jeffrey M. Roseman, MD, PhD, MPH,Holly M. Bastian, MD, MSPH, Barri J. Fessler, MD, Gerald McGwin,Jr., MS, PhD, Martha L. Sanchez, MD, MPH: The University ofAlabama at Birmingham; 2Luis M. Vila, MD: The University of PuertoRico Medical Sciences Campus, San Juan, Puerto Rico; 3John D.Reveille, MD: The University of Texas Health Science Center atHouston.

Current investigators and staff in the LUMINA (LUpus inMInorities, NAture versus nurture) study are as follows: Graciela S.Alarcon, MD, MPH, Holly M. Bastian, MD, MSPH, Barri J. Fessler,MD, Gerald McGwin, Jr., PhD, MS, Jeffrey Roseman, MD, PhD,MPH, Ana Bertoli, MD, Monica Fernandez, MD, Martha L. Sanchez,MD, MPH, Ellen Sowell (University of Alabama at Birmingham);John D. Reveille, MD, Chul Ahn, PhD, Robert Sandoval, BA, BinhVu, BS (University of Texas Health Science Center at Houston); LuisVila, MD, William Borges, AA, Carmine Pinilla BS (University ofPuerto Rico Medical Sciences Campus).

Address correspondence and reprint requests to Graciela S.Alarcon, MD, MPH, 830 Faculty Office Tower, 510 20th Street South,Birmingham, AL 35294. E-mail: [email protected].

Submitted for publication December 2, 2004; accepted inrevised form February 18, 2005.

1655

more frequent in postmenopausal women. Age, ratherthan menopausal status, independently contributed todamage accrual, renal involvement, and vascular arte-rial events in these women.

Conclusion. A hypoestrogenemic state secondaryto menopause appears not to be protective againstdisease activity and damage accrual. Age rather thanmenopausal status is a strong independent predictor ofdamage accrual and of vascular events in women withlupus.

Systemic lupus erythematosus (SLE) is an auto-immune disease of unknown etiology that affects womenmore frequently, particularly during their reproductiveyears (female to male ratio 9:1) (1–3). Several geneticand nongenetic factors have been implicated in itspathogenesis, course, and outcome. Menses, meno-pause, and pregnancy are thought to modify the courseof the disease. For example, sex hormones have beenconsidered as disease modulators (4,5), although theexact mechanism of these effects is not well understood;however, estrogens can stimulate B lymphocytes andsuppress cell-mediated immunity (6,7). Sex hormoneshave also been implicated in apoptosis of peripheralblood mononuclear cells with reduced production oftumor necrosis factor �, which may allow survival ofautoimmune cells in SLE patients (8,9). Several murinemodels of SLE have shown an association between sexhormones and autoimmunity (10,11); for example, fe-male (NZB � NZW)F1 mice develop autoantibodies tonuclear components, immune complex glomerulone-phritis, and arthritis.

When the premenopausal and postmenopausaltime periods have been compared in women with SLE, amodest decrease in disease activity after natural meno-pause has been observed (3). Similar observations havebeen documented in patients with ovarian failure sec-ondary to cyclophosphamide therapy; these patientsseem to experience less frequent and less severe diseaseflares than those cyclophosphamide-treated patientswho did not undergo secondary ovarian failure (1).Finally, the frequency of SLE flares appears to decreasein women as they become postmenopausal (2). How-ever, decreased disease activity over the duration of thedisease has been observed in all SLE patients, so theeffect of menopause could simply be due to age anddisease duration.

The objectives of this study were to compare theclinical manifestations, disease course, disease activity,damage accrual, vascular events, and medication use in

women with lupus from a multiethnic US cohort, accord-ing to disease onset prior to versus after menopause.

PATIENTS AND METHODS

As previously described (12), LUMINA (LUpus inMInorities, NAture versus nurture) is a longitudinal study ofoutcome in lupus. The current analyses are limited to thewomen in the cohort. All LUMINA patients met the AmericanCollege of Rheumatology (ACR) criteria for the classificationof SLE (13), had disease duration �5 years, were �16 years ofage, were of defined ethnicity (African American, Hispanic[defined as Mexican and Central American (Texas) or PuertoRican ancestry], or Caucasian), and were living in the geo-graphic recruitment area of the participating centers (theUniversity of Alabama at Birmingham, the University of TexasHealth Science Center at Houston, or the University of PuertoRico Medical Sciences Campus). Prior to enrollment, allmedical records were reviewed to confirm patient eligibility,and to gather socioeconomic/demographic information andrelevant clinical data from the time of diagnosis to enrollment.Every patient has a baseline visit (time 0); followup visits areconducted every 6 months for the first year, and yearlythereafter. A LUMINA study visit consists of an interview, aphysical examination, and laboratory tests. Data for missedstudy visits are obtained, whenever possible, by review of allavailable medical records. The time of diagnosis was defined asthe time at which patients met 4 components of the ACRcriteria for SLE. Disease duration was defined as the intervalbetween time of diagnosis and time 0.

Menopause was defined at disease onset as self-reported climacteric symptoms (hot flashes), and/or amenor-rhea lasting �6 months, and/or bilateral oophorectomy (withor without hysterectomy), and/or increased follicle-stimulatinghormone (FSH) values above the upper limit of normal forreproductive-age women, and/or hormone replacement ther-apy (HRT) use. Premature gonadal failure was defined assecondary amenorrhea prior to age 40 years, regardless of itscause. Women were categorized into pre- and postmenopausegroups; those who experienced premature gonadal failure wereincluded in the latter category.

Variables. As previously reported (14), the LUMINAdatabase includes variables from the following domains:socioeconomic/demographic, clinical, immunologic, immuno-genetic, behavioral, and psychological. These variables aremeasured at time 0 and at every subsequent visit. Only thevariables included in these analyses will be described.

Variables from the socioeconomic/demographic do-main included age, ethnicity, education, poverty (as defined bythe US Federal government adjusted for the number ofsubjects in the household) (15), health insurance, and smoking.Clinical variables included the number of ACR criteria at timeof diagnosis, disease duration from time of diagnosis tobaseline visit, followup time from baseline visit to last visit,total disease duration from diagnosis to last visit, weight,height, body mass index (BMI), disease activity and damage,disease manifestations, laboratory results, and medications.

Disease activity was assessed using the Systemic LupusActivity Measure (SLAM) (16) at time 0 and at last visit; theaverage SLAM value of all recorded scores (from time of

1656 FERNANDEZ ET AL

diagnosis to last visit) was calculated as a measure of diseaseactivity over time. The SLAM is a validated instrument, whichincludes 23 clinical manifestations and 7 laboratory parametersthat must be present some time during the preceding monthand attributable to SLE.

Damage was measured with the Systemic Lupus Inter-national Collaborating Clinics/American College of Rheuma-tology damage index (SDI) (17) at time 0 and last visit. Thisindex documents cumulative and irreversible damage, irrespec-tive of its cause, in 12 different organ systems. To be scored,each manifestation must be present for at least 6 months,unless otherwise noted in the instructions accompanying thisinstrument.

Cumulative organ system manifestations secondary tolupus were recorded individually and by categories as follows:integument involvement (malar rash, discoid rash, photosensi-tivity rash, nailfold infarction, purpura, other vasculitic lesions,Raynaud’s phenomenon, and/or mucosal ulcers); musculoskel-etal involvement (arthralgias, arthritis, osteoporosis, myalgias,and/or myositis); neurologic involvement (seizures, psychosis,headaches, cerebrovascular accidents, neuropathies, transversemyelitis, myelopathy, and/or radiculopathy); renal involvement(nephritis by histopathology [World Health Organization classII or higher] and/or 2–3� proteinuria); and hematologicinvolvement (hemolytic anemia, leukopenia, lymphopenia,and/or thrombocytopenia). Vascular arterial events were re-corded as present if myocardial infarction, definite or classicangina, and/or a vascular procedure for myocardial infarction(coronary artery bypass graft), stroke, intermittent claudica-tion, and/or peripheral arterial thrombosis were present; ve-nous thrombotic events were recorded as present if peripheraland/or visceral venous thrombosis were documented. Preva-lent and incident cases occurring after time of diagnosis wereincluded.

Cumulative exposure to glucocorticoids, hydroxychlo-roquine, cyclophosphamide, azathioprine, and methotrexatewas also recorded. Exposure to HRT was recorded as presentif HRT was used any time after menopause, regardless of itsduration. The average glucocorticoid dose was calculated inmg/day.

The following laboratory variables were recorded attime 0: levels of total cholesterol (normal �200 mg/dl), low-density lipoprotein (LDL) cholesterol (calculated using the

Friedewald formula [normal �130 mg/dl]), high-density li-poprotein (HDL) cholesterol (normal �35 mg/dl), and triglyc-erides (normal �205 mg/dl). Serum C-reactive protein (CRP),measured by high-sensitivity immunometric assay (hsCRP)(Immulite 2000 Diagnostic Products, Los Angeles, CA), wasalso recorded.

Samples were analyzed for the presence of autoanti-bodies, including antinuclear antibodies (by immunofluores-cence using HEp-2 cell line), anti–double-stranded DNA(anti-dsDNA) (by immunofluorescence against Crithidia luci-liae [normal �1:10]), anti-Sm and anti-Ro (by counterimmu-noelectrophoresis against human spleen and calf thymus ex-tract) (18,19), IgG and IgM antiphospholipid antibodies (aPL)(abnormal �13 IgG phospholipid units/ml and/or �13 IgMphospholipid units/ml, by enzyme-linked immunoabsorbentassay) (20), and lupus anticoagulant (LAC) (Staclot test;Diagnostica Stago 92600, Asnieres-Sur-Seine, France) (21).Patients were considered to have aPL if they had IgM and/orIgG aPL and/or LAC.

Statistical analysis. Features from the different do-mains were compared between the premenopausal and thepostmenopausal women using standard statistical tests. Stu-dent’s t-test and the chi-square test were carried out forcontinuous and categorical variables, respectively. For pre-menopausal patients who became menopausal, last visit wasthe visit prior to the occurrence of menopause. When the SDIscore was compared, women with premature gonadal failurewere excluded (n � 7), so this specific damage item would notcount and the data would not be biased toward more damagein the postmenopausal group. Analyses regarding the SDIwere done including and excluding patients with osteoporosis,fractures, or vertebral collapse (all expected to increase withage). P values less than or equal to 0.05 were consideredsignificant.

Multivariable analyses (linear and logistic regressionsusing backward selection method) were then performed, toexamine the role and association of menopause with specificoutcomes or events based on the univariable analyses (damageaccrual, disease manifestations, vascular arterial, and venousthrombotic events). Adjustment for possible confounders wasperformed; age, menopause, and ethnicity were entered in allmodels. All statistical analyses were performed using the SPSSprogram version 10.0 (SPSS, Chicago, IL).

Table 1. Salient socioeconomic and demographic features of the LUMINA patients as a function ofmenopausal status*

FeaturePremenopausal

(n � 436)Postmenopausal

(n � 82) P

Age, mean � SD years 33.4 � 9.9 52.6 � 10.4 �0.001Ethnicity 0.004

Hispanic from Texas 21 13Hispanic from Puerto Rico 18 15African American 38 29Caucasian 23 42

Education, mean � SD years 13.0 � 9.9 12.6 � 3.4 0.260Poverty, below US Federal government guidelines 36 29 0.303Health insurance 78 88 0.037Smoker 12 15 0.588

* Except where indicated otherwise, values are percents. LUMINA � LUpus in MInorities, NAture versusnurture.

INFLUENCE OF MENOPAUSE IN LUMINA PATIENTS 1657

RESULTS

There were 518 women in the LUMINA cohortat the time these analyses were performed; 436 (84.2%)were premenopausal and 82 (15.8%) were postmeno-pausal at disease onset. Thirty postmenopausal womenwere taking HRT.

Univariable analyses. Salient socioeconomic de-mographic, and clinical features of these women areshown in Table 1 and Table 2. As expected, the pre-menopausal women were younger (mean 33.4 years)than the postmenopausal women (mean 52.6 years).Within the postmenopausal group there was a high

proportion of Caucasian patients (42%), whereas thepremenopausal group had a high proportion of AfricanAmericans (38%) (P � 0.004). A higher proportion ofpostmenopausal women than premenopausal womenhad health insurance (88% versus 78%; P � 0.037). Allother clinical and socioeconomic/demographic featureswere comparable between the pre- and postmenopausalpatients with the exception of damage, both at time 0and last visit, which was higher among the postmeno-pausal than the premenopausal women (mean � SD attime 0 1.1 � 1.5 versus 0.6 � 1.0 [P � 0.001]; at last visit2.2 � 2.5 versus 1.2 � 1.8 [P � 0.001]). The difference in

Table 2. Selected clinical features of the LUMINA patients as a function of menopausal status*



(n � 82) P

Body mass index, kg/m2 26.7 � 8.6 27.5 � 5.0 0.511Disease duration, months† 17.8 � 16.5 20.3 � 17.3 0.214Followup time, months‡ 35.0 � 32.9 35.7 � 33.7 0.860Total disease duration, months§ 53.0 � 38.4 56.1 � 42.7 0.514ACR criteria, number present 5.5 � 1.3 5.2 � 1.1 0.085SLAM score

Baseline 9.7 � 5.8 8.6 � 4.8 0.124Last visit 8.2 � 5.5 8.0 � 5.0 0.690Average 7.7 � 4.7 7.2 � 3.7 0.343

SDI scoreBaseline 0.6 � 1.0 1.1 � 1.5 �0.001Last visit 1.2 � 1.8 2.2 � 2.5 �0.001Difference between visits 0.6 � 1.4 1.0 � 1.7 0.044

* Values are the mean � SD. LUMINA � LUpus in MInorities, NAture versus nurture; ACR �American College of Rheumatology; SLAM � Systemic Lupus Activity Measure; SDI � Systemic LupusInternational Collaborating Clinics damage index.† From time of diagnosis to enrollment or baseline visit.‡ From baseline visit to last visit.§ From time of diagnosis to last visit.

Table 3. Cumulative clinical features in the LUMINA patients by organ system as a function ofmenopausal status*

FeaturesPremenopausal


(n � 82) P

Integument involvement 94 96 0.604Raynaud’s phenomenon 56 45 0.070

Musculoskeletal involvement 84 88 0.505Neurologic involvement 57 61 0.543Hematologic involvement 78 72 0.256Renal involvement 48 28 0.001

Proteinuria 2� 32 24 0.016Proteinuria 3� 41 25 0.006Membranous glomerulonephritis 11 4 0.042

Vascular arterial events† 7 17 0.004Venous thrombotic events‡ 7 13 0.080

* Values are percents. LUMINA � LUpus in MInorities, NAture versus nurture.† Myocardial infarction, angina, stroke, intermittent claudication, and/or peripheral arterial thrombosis.‡ Peripheral and/or visceral thrombosis.


damage accrual between time 0 and last visit was alsohigher among the postmenopausal than the premeno-pausal women (1.0 � 1.7 versus 0.6 � 1.4; P � 0.044).

In contrast, and as shown in Table 3, there weresome differences in disease manifestations between the2 groups of women. For example, vascular arterial eventswere more frequent in the postmenopausal than in thepremenopausal women (17% versus 7%; P � 0.004); thesame was the case for venous thrombotic events (13%versus 7%), but the difference was not statisticallysignificant (P � 0.080). In contrast, severe proteinuriaand membranous glomerulonephritis were more com-mon in the premenopausal patients (41% versus 25%[P � 0.006] and 11% versus 4% [P � 0.042], respec-tively), as was renal involvement in general (48% in thepremenopausal and 28% in the postmenopausal women;P � 0.001). Raynaud’s phenomenon was also more fre-quent in the premenopausal women (56%) than in thepostmenopausal women (45%), but the difference did notreach statistical significance (P � 0.070). All other diseasemanifestations were comparable between the 2 groups.

Medication use (glucocorticoids, methotrexate,cyclophosphamide, and azathioprine) was comparable inthe 2 patient groups (data not shown). The only medi-cation whose use differed in the 2 groups was statins(higher in the postmenopausal than in the premeno-pausal women [9% versus 3%; P � 0.014]). No signifi-

cant differences were observed between the 2 groups(data not shown) with regard to the baseline autoanti-bodies, except that anti-dsDNA antibodies were morefrequently positive in premenopausal women (31%)than in postmenopausal women (18%) (P � 0.032).

The proportion of patients with elevated LDLlevels was higher among postmenopausal women thanpremenopausal women (32% versus 21%, P � 0.046).Postmenopausal women also showed higher total cho-lesterol levels than premenopausal patients (mean 183.0mg/dl versus 167.4 mg/dl; P � 0.048). In contrast, themean values of hsCRP were in the normal range andquite similar in the 2 groups (12.5 mg/liter in premeno-pausal women versus 14.9 mg/liter in postmenopausalwomen). These data are shown in Table 4.

Multivariable analyses. Disease manifestations.Given the findings in the univariable analyses, theimpact of menopause on the occurrence of renal in-volvement in general was examined. We adjusted forpossible confounder variables such as age, ethnicity,disease duration, and anti-dsDNA in this regression. Inthese analyses, menopause was not found to be aprotective factor against the occurrence of renal involve-ment; in contrast, adjustment variables emerged aseither being risk factors for or protective against such anoccurrence. Anti-dsDNA antibodies (odds ratio [OR]2.30, 95% confidence interval [95% CI] 1.38–3.84, P �

Table 4. Baseline laboratory features of the LUMINA patients as a function of menopausal status*



(n � 82) P

hsCRP, mean � SD mg/liter 12.5 � 29.4 14.9 � 30.6 0.533Total cholesterol, mean � SD mg/dl 167.4 � 58.7 183.0 � 75.4 0.048High LDL cholesterol (�130 mg/dl) 21 32 0.046Low HDL cholesterol (�35 mg/dl) 49 59 0.128High triglycerides (�205 mg/dl) 14 19 0.280

* Except where indicated otherwise, values the percent of patients with abnormal measurements.LUMINA � lupus in minorities, nature versus nurture; hsCRP � C-reactive protein measured byhigh-sensitivity immunometric assay; LDL � low-density lipoprotein; HDL � high-density lipoprotein.

Table 5. Multivariable logistic regression analyses of renal involvement*

Feature Standardized coefficient Odds ratio (95% CI) P

Age �0.034 0.96 (0.94–0.99) 0.006Ethnicity†

Hispanic (Puerto Rico) �1.706 0.18 (0.07–0.46) �0.001African American 0.650 1.91 (1.07–3.41) 0.028Caucasian �1.199 0.30 (0.15–0.57) �0.001

Menopause 0.018 1.01 (0.47–2.18) 0.964Anti-dsDNA antibodies 0.836 2.30 (1.38–3.84) �0.001

* 95% CI � 95% confidence interval; anti-dsDNA � anti–double-stranded DNA.† The Hispanic (Texas) group is the reference group.


0.001) were independently associated with renal involve-ment, whereas older age (OR 0.96, 95% CI 0.94–0.99,P � 0.006) emerged as a protective factor against theoccurrence of renal involvement. African Americanethnicity appeared to be a risk factor for the occurrenceof renal involvement (OR 1.91, 95% CI 1.07–3.41, P �0.028) when Hispanic ethnicity (from Texas) was thereference category; being of Caucasian or Hispanic(Puerto Rico) ethnicity was protective against the occur-rence of renal involvement (OR 0.30, 95% CI 0.15–0.57,P � 0.001 for Caucasians; OR 0.18, 95% CI 0.072–0.459,P � 0.001 for Hispanics from Puerto Rico). These dataare shown in Table 5. When Caucasians were consideredthe reference group, the results were consistent with theones presented.

Damage. Data on damage accrual are depicted inTable 6. Menopausal status was not associated with theoccurrence of damage at time 0. In contrast, some of theadjustment variables in these analyses were indepen-dently associated with damage accrual. At time 0, thesevariables were age (P � 0.038), African Americanethnicity (P � 0.008), and disease duration (P � 0.015).Elevated levels of serum HDL cholesterol showed anegative association with damage accrual (P � 0.028).When age was not entered into the model, menopauseemerged as a contributing factor for the occurrence ofdamage at time 0 (P � 0.003 [data not shown]). Aninteraction between age and menopause could not bedemonstrated.

Menopausal status was also not associated with

Table 6. Multivariable linear regression analyses of damage accrual at time 0 and last visit

Feature

Damage at first visit Damage at last visit

Standardizedcoefficient t P

Standardizedcoefficient t P

Age 0.126 2.087 0.038 0.135 2.455 0.015Ethnicity*

Hispanic from Texas 0.034 0.590 0.556 0.117 2.106 0.036Hispanic from Puerto Rico �0.082 �1.461 0.145 �0.015 �0.278 0.781African American 0.156 2.686 0.008 0.165 2.838 0.005

Disease duration† 0.116 2.434 0.015Total disease duration‡ 0.196 4.196 �0.001Menopause 0.061 1.030 0.304 0.080 1.309 0.191Renal involvement 0.143 2.786 0.006Neurologic involvement§ 0.159 3.502 0.001HDL cholesterol¶ �0.108 �2.202 0.028

* The Caucasian group is the reference group for both models.† From time of diagnosis to time 0.‡ From time of diagnosis to last visit.§ Excludes psychosis and seizures.¶ HDL � high-density lipoprotein.

Table 7. Multivariable logistic regression analyses of vascular events*

Feature

Arterial vascular events Venous thrombosis

Standardizedcoefficient

Odds ratio(95% CI) P

Standardizedcoefficient

Odds ratio(95% CI) P

Age 0.046 1.04 (1.00–1.08) 0.019 0.035 1.03 (0.99–1.07) 0.078Smoker 1.166 3.21 (1.29–7.97) 0.012 1.072 2.92 (1.18–7.20) 0.020Total disease duration† 0.018 1.01 (1.00–1.02) �0.001 0.002 1.00 (0.99–1.01) 0.717Menopause 0.872 2.39 (0.68–8.06) 0.159 0.660 1.93 (0.56–6.58) 0.291Mean glucocorticoid dose‡ 0.042 1.04 (1.00–1.08) 0.025Hormone replacement therapy �1.795 0.16 (0.03–1.01) 0.051 0.880 0.41 (0.07–2.19) 0.415

* Vascular events were defined as either arterial vascular events or venous thrombosis. Arterial vascular events included myocardial infarction,angina, stroke, intermittent claudication, and/or peripheral arterial thrombosis. Venous thrombosis was defined as peripheral and/or visceral. 95%CI � 95% confidence interval.† From time of diagnosis to last visit.‡ Not adjusted for venous thrombosis model.


the occurrence of damage accrual at last visit. Possibleconfounding variables that were adjusted for in thismodel, such as age (P � 0.015), total disease duration(P � 0.001), renal involvement (P � 0.006), and neuro-logic manifestations (P � 0.001), were independentlyassociated with damage accrual at last visit. Being ofHispanic (from Texas) or of African American ethnicitywas also predictive of damage accrual at last visit (P �0.036 and P � 0.005, respectively), when compared withpatients of Caucasian ethnicity. These damage analyseswere also done excluding those patients who had osteo-porosis, fractures, or vertebral collapse; results werecomparable (data not shown).

Vascular arterial and venous thrombotic events.Menopausal status was not a predictive factor for theoccurrence of vascular arterial or venous thromboticevents. These data are depicted in Table 7. Vasculararterial events were independently associated with someof the variables adjusted for in the model; these were age(OR 1.04, 95% CI 1.00–1.08, P � 0.019), smoking (OR3.21, 95% CI 1.29–7.97, P � 0.012), and disease-relatedfactors such as total disease duration (OR 1.01, 95% CI1.00–1.02, P � 0.001) and the mean dose of glucocorti-coids (OR 1.04, 95% CI 1.04–1.08, P � 0.025). HRT usewas, on the other hand, negatively associated with theoccurrence of these events, but the results were only ofborderline significance (OR 0.16, 95% CI 0.027–1.011,P � 0.051). Venous thrombosis was associated only withsmoking (OR 2.92, 95% CI 1.185–7.196, P � 0.020).

DISCUSSION

In order to further characterize the differences indisease expression and outcome in patients with SLE asa function of their menopausal status at disease onset,we have taken advantage of a large, multiethnic, longi-tudinal cohort of patients, the LUMINA cohort. Giventhat in LUMINA, Caucasian patients have been older atenrollment (14), we were not surprised to find thatCaucasian women were overrepresented in the post-menopausal group; this finding, however, has not beenpreviously reported in the literature. Some disease man-ifestations differed in the 2 patient groups. Some of thedifferences can be considered a function of age or of ageand menopause; for example, postmenopausal womentended to experience more vascular arterial events andlipid abnormalities. Renal involvement was more fre-quent overall in the premenopausal women, as was thepresence of Raynaud’s phenomenon. Despite these dif-ferences, disease activity, whether at diagnosis, at enroll-ment, or over time, appears to be of comparable mag-

nitude in both patient groups. Finally, even afterexcluding women who had early menopause due tosecondary ovarian failure, damage accrual at time 0 or atlast visit was higher in the postmenopausal women.However, age, but not menopause, appeared to be thedecisive factor, as demonstrated in the multivariableanalyses.

The relationship between age, disease severity,and ethnicity is unclear, but it is possible that age, ratherthan menopause, exerts a modulatory effect on SLE(14,22). This seems to be also evident in terms of renalinvolvement, which was more frequent among the pre-menopausal women. Again this relates to age ratherthan menopause, as we (23) and others (24,25) havedemonstrated previously.

The frequency of anti-Ro and anti-La positivityhas been reported to increase with age (26,27). We couldnot validate this association in our patients, because thedata were not longitudinal: anti-Ro and anti-La statuswas measured only at time 0. On the other hand,anti-dsDNA antibodies were more common in the pre-menopausal women, which probably relates to the morefrequent occurrence of renal involvement among thisgroup of patients (27,28).

Although hyperlipidemia has been reported inpatients with SLE during periods of disease activity(29,30), it is unlikely to be the explanation for theincreased lipid levels we found in our postmenopausalpatients, since disease activity was comparable in bothgroups of women. Given that hyperlipidemia that per-sists once lupus disease activity declines should betreated (with statins), it is not surprising that we foundan increased use of statins among the postmenopausalwomen; these data clearly reflect confounding by indi-cation (31), since the postmenopausal women were morelikely not only to have hyperlipidemia, but to haveexperienced more vascular events.

The relationship between disease activity, diseasecourse, and sex hormones in SLE is controversial; how-ever, the association between a hypoestrogenemic stateearly in the disease (for example, due to cyclo-phosphamide-induced ovarian failure) and a more be-nign disease course has been reported by several inves-tigators (1–3,5). Sanchez-Guerrero and collaborators,for example, reported on a cohort of 30 women beforeand after they became menopausal; they found a modestdecrease in disease activity (measured with the SLEDisease Activity Index) as women underwent naturalmenopause (3). Likewise, Mok and collaborators stud-ied a group of women with SLE whose disease startedafter menopause (n � 17) and another group whose


disease started before menopause, but who becamemenopausal during the study (n � 34). They found thatlupus flares decreased in frequency and severity aswomen become menopausal (2). In still another study,Mok et al studied 54 premenopausal women who re-ceived treatment with cyclophosphamide; 14 underwentovarian failure and experienced disease flares less fre-quently than the 40 women who were still menstruating(1). The authors concluded that hypoestrogenemia ac-counted for a decline in disease activity.

These findings are in contrast to the data we havepresented; although we did not ascertain disease flares,disease activity was comparable in both groups, as weredisease-specific treatments, including glucocorticoidsand immunosuppressive agents that could amelioratedisease activity and can be regarded as proxies fordisease activity. We do not have an exact explanation forthe discrepancy between our data and those from theliterature, although methodologic differences may bethe underlying factor.

Another possible explanation for the discrepantresults between our study and those cited may relate tothe fact that we studied one group of women whoselupus began before menopause and another groupwhose lupus began after, rather than the same womenwhose disease started during their premenopausal yearsand who were followed up after they became meno-pausal. An abnormal pathway of estrogen metabolismhas been recognized in patients with SLE for decades(32) (elevated serum concentrations of 16�-hydroxylatedestrogens, a mitogenic tumor growth-stimulating factor[33]). However, it appears that the conversion of 17�-estradiol either to 16�-hydroxylated estrogens or to2-hydroxylated estrogens (growth-inhibitor factor) oc-curs peripherally (34); thus, menopause may have littleinfluence on disease manifestations and course, as ourdata seem to indicate.

The clinical effect of sex hormones and meno-pause is not limited to SLE; it occurs in other rheumaticdiseases as well. The possible effect of sex hormones onthe pathogenesis of rheumathoid arthritis (RA), forexample, is suggested by the characteristic age and sexdistribution of the disease (35) and worsening of diseaseactivity related to changes in FSH or estrogen levels withoophorectomy (36) or menopause (37) or improvementwith androgens (38); furthermore, remissions occur inapproximately two-thirds of women with RA who be-come pregnant (39). As in patients with SLE, elevatedserum concentrations of 16�-hydroxylated estrogenshave been found in the synovial fluid of patients with RA(40). With regard to scleroderma, Italian investigators

have reported an association between postmenopausalstatus (alone or in combination with HLA–B35 andCREST syndrome [calcinosis, Raynaud’s phenomenon,esophageal dysmotility, intestinal dymostility, sclerodac-tyly, and telangiectasias]) and the occurrence of pulmo-nary hypertension (41). Finally, sicca symptoms in pa-tients with Sjogren’s syndrome have also been related tochanges in female sex steroids (42).

As noted above, the differences found regardingdamage accrual and renal involvement seem to relate toage rather than to menopause, as demostrated in themultivariable analyses. Of note, patients with ovarianfailure as documented by the SDI were excluded fromthese analyses. The analyses were also repeated exclud-ing patients with osteoporosis, but the results remainedunchanged. When age and menopause were included inthe models as an interaction term, age remained as anindependent risk factor for damage accrual.

Finally, given that age is a known risk factor forthe occurrence of vascular events in the general popu-lation, and in SLE patients in particular (43,44), wewanted to determine if menopausal status, as a hy-poestrogenemic state, would be the underlying factorsupporting the role of age in the occurrence of vascularevents. In fact, that was not the case; age, but notmenopausal status, was an independent predictor ofvascular arterial events. However, HRT use appeared tobe a protective factor for the development of arterialvascular events. We realize that our study was notdesigned to examine the role of HRT and that its use hasdecreased significantly after the data from the Women’sHealth Initiative study were published (45); however,some beneficial effects of HRT use include higher levelsof HDL cholesterol, decreased LDL cholesterol (46,47),and decreased plasminogen activator inhibitor type 1(48). Given the relatively young age of our postmeno-pausal patients, we are not at all surprised that HRTappeared to be protective, as such has been implied inthe meta-analysis by Salpeter et al (49).

We acknowledge that our study is not withoutlimitations. First, one of the definitions of menopausewas amenorrhea of �6 months’, rather than �12 months’,duration, which may have led to some misclassification;however, if some premenopausal patients were classifiedas postmenopausal, we would expect their data to havediminished rather than accentuated the differences ob-served. Second, menopausal status was not corroboratedby measurement of FSH or estradiol values in allpatients as menses ceased. The menopause definition wehave used is, however, consistent with that documentedin the gynecology literature (50), and more stringent


than the one included in the SDI, which only requiresthe presence of premature gonadal failure prior to age40 years (17,51). Third, damage accrual and vascularevents were ascertained from the available clinical data,rather than systematically, using a defined protocol.Fourth, we did not ascertain disease flares, since we onlyassessed disease activity (with the SLAM) during studyvisits (every 6 months for the first year, and yearlythereafter). Instead, we used the average SLAM score,although this is not equivalent to flares.

Despite these limitations, our research is basedon the largest multiethnic postmenopausal group ofpatients for whom data have been obtained longitudi-nally following a pre-established protocol; moreover, thedata presented, while not supporting the role of meno-pause in either disease activity, disease manifestations,damage accrual, or thrombotic events, nevertheless areconsistent with data previously generated from ourcohort, strongly supporting the validity of this study. Wetherefore conclude that age, rather than menopause, is astrong predictor of damage accrual, vascular arterialevents, and venous thrombotic events in women withSLE whose disease starts after menopause.

REFERENCES

1. Mok CC, Wong RW, Lau CS. Ovarian failure and flares ofsystemic lupus erythematosus. Arthritis Rheum 1999;42:1274–80.

2. Mok CC, Lau CS, Ho CT, Wong RW. Do flares of systemic lupuserythematosus decline after menopause? Scand J Rheumatol1999;28:357–62.

3. Sanchez-Guerrero J, Villegas A, Mendoza-Fuentes A, Romero-Diaz J, Moreno-Coutino G, Cravioto MC. Disease activity duringthe premenopausal and postmenopausal periods in women withsystemic lupus erythematosus. Am J Med 2001;111:464–8.

4. Wen C, Li LS. Blood levels of sex hormone in lupus nephritis andtheir relationship to lupus activity. Chin Med J (Engl) 1993;106:49–52.

5. Lahita RG. Sex hormones and systemic lupus erythematosus.Rheum Dis Clin North Am 2000;26:951–68.

6. Bynoe MS, Grimaldi CM, Diamond B. Estrogen up-regulatesBcl-2 and blocks tolerance induction of naive B cells. Proc NatlAcad Sci U S A 2000;97:2703–8.

7. Tsokos GC, Wong HK, Enyedy EJ, Nambiar MP. Immune cellsignaling in lupus. Curr Opin Rheumatol 2000;12:355–63.

8. Lahita RG. The importance of estrogens in systemic lupus ery-thematosus. Clin Immunol Immunopathol 1992;63:17–8.

9. Evans MJ, MacLaughlin S, Marvin RD, Abdou NI. Estrogendecreases in vitro apoptosis of peripheral blood mononuclear cellsfrom women with normal menstrual cycles and decreases TNF-�production in SLE but not in normal cultures. Clin ImmunolImmunopathol 1997;82:258–62.

10. Roubinian JR, Talal N, Greenspan JS, Goodman JR, Siiteri PK.Effect of castration and sex hormone treatment on survival, antinucleic acid antibodies, and glomerulonephritis in NZB/NZW F1mice. J Exp Med 1978;147:1568–83.

11. Yoshida S, Castles JJ, Gershwin ME. The pathogenesis of auto-immunity in New Zealand mice. Semin Arthritis Rheum 1990;19:224–42.

12. Alarcon GS, Friedman AW, Straaton KV, Moulds JM, Lisse J,Bastian HM, et al. Systemic lupus erythematosus in three ethnicgroups. III. A comparison of characteristics early in the naturalhistory of the LUMINA cohort. Lupus 1999;8:197–209.

13. Hochberg MC. Updating the American College of Rheumatologyrevised criteria for the classification of systemic lupus erythema-tosus [letter]. Arthritis Rheum 1997;40:1725.

14. Reveille JD, Moulds JM, Ahn C, Friedman AW, Baethge B,Roseman J, et al, for the LUMINA Study Group. Systemic lupuserythematosus in three ethnic groups. I. The effects of HLA classII, C4, and CR1 alleles, socioeconomic factors, and ethnicity atdisease onset. Arthritis Rheum 1998;41:1161–72.

15. US Department of Commerce, Bureau of the Census. Currentpopulation reports, series P-23, no. 28 and series P-60, no. 68 andsubsequent years. Washington, DC: Housing and HouseholdEconomic Statistics Division; 1995.

16. Gladman DD, Goldsmith CH, Urowitz MB, Bacon P, BombardierC, Isenberg D, et al. Sensitivity to change of 3 systemic lupuserythematosus disease activity indices: international validation.J Rheumatol 1994;21:1468–71.

17. Gladman DD, Urowitz MB, Goldsmith CH, Fortin P, Ginzler E,Gordon C, et al. The reliability of the Systemic Lupus Interna-tional Collaborating Clinics/American College of RheumatologyDamage Index in patients with systemic lupus erythematosus.Arthritis Rheum 1997;40:809–13.

18. Aarden LA, de Groot ER, Feltkamp TE. Immunology of DNA.III. Crithidia luciliae, a simple substrate for the determination ofanti-dsDNA with the immunofluorescence technique. Ann N YAcad Sci 1975;254:505–15.

19. Hamilton RG, Harley JB, Bias WB, Roebber M, Reichlin M,Hochberg MC, et al. Two Ro (SS-A) autoantibody responses insystemic lupus erythematosus: correlation of HLA–DR/DQ spec-ificities with quantitative expression of Ro (SS-A) autoantibody.Arthritis Rheum 1988;31:496–505.

20. Harris EN. Special report: the second international anti-cardiolipin standardization workshop/the Kingston Anti-Phospho-lipid Antibody Study (KAPS) Group. Am J Clin Pathol 1990;94:476–84.

21. Triplett DA, Barna LK, Unger GA. A hexagonal (II) phasephospholipid neutralization assay for lupus anticoagulant identifi-cation. Thromb Haemost 1993;70:787–93.

22. McCarty DJ, Manzi S, Medsger TA Jr, Ramsey-Goldman R,LaPorte RE, Kwoh CK. Incidence of systemic lupus erythemato-sus: race and gender differences. Arthritis Rheum 1995;38:1260–70.

23. Bastian HM, Roseman JM, McGwin G Jr, Alarcon GS, FriedmanAW, Fessler BJ, et al. Systemic lupus erythematosus in threeethnic groups. XII. Risk factors for lupus nephritis after diagnosis.Lupus 2002;11:152–60.

24. Alba P, Bento L, Cuadrado MJ, Karim Y, Tungekar MF, Abbs I,et al. Anti-dsDNA, anti-Sm antibodies, and the lupus anticoagu-lant: significant factors associated with lupus nephritis. AnnRheum Dis 2003;62:556–60.

25. Blatt NB, Glick GD. Anti-DNA autoantibodies and systemic lupuserythematosus. Pharmacol Ther 1999;83:125–39.

26. Cervera R, Khamashta MA, Font J, Sebastiani GD, Gil A, LavillaP, et al, and The European Working Party on Systemic LupusErythematosus. Systemic lupus erythematosus: clinical and immu-nologic patterns of disease expression in a cohort of 1,000 patients.Medicine (Baltimore) 1993;72:113–24.

27. Jonsson H, Nived O, Sturfelt G. The effect of age on clinical andserological manifestations in unselected patients with systemiclupus erythematosus. J Rheumatol 1988;15:505–9.

28. Wilson HA, Hamilton ME, Spyker DA, Brunner CM, O’BrienWM, Davis JS IV, et al. Age influences the clinical and serologicexpression of systemic lupus erythematosus. Arthritis Rheum1981;24:1230–5.


29. Hahn BH. Systemic lupus erythematosus and accelerated athero-sclerosis. N Engl J Med 2003;349:2379–80.

30. Molenaar ET, Voskuyl AE, Dinant HJ, Bezemer PD, Boers M,Dijkmans BA. Progression of radiologic damage in patients withrheumatoid arthritis in clinical remission. Arthritis Rheum 2004;50:36–42.

31. Rubin DB. Estimating causal effects from large data sets usingpropensity scores. Ann Intern Med 1997;127:757–63.

32. Lahita RG, Bradlow HL, Kunkel HG, Fishman J. Alterations ofestrogen metabolism in systemic lupus erythematosus. ArthritisRheum 1979;22:1195–8.

33. Vandewalle B, Lefebrre J. Opposite effects of estrogen andcatecholestrogen on hormone-sensitive breast cancer cell growthand differentiation. Mol Cell Endocrinol 1989;61:239–46.

34. Weidler C, Harle P, Schedel J, Schmidt M, Scholmerich J, StraubRH. Patients with rheumatoid arthritis and systemic lupus ery-thematosus have increased renal excretion of mitogenic estrogensin relation to endogenous antiestrogens. J Rheumatol 2004;31:489–94.

35. Goemaere S, Ackerman C, Goethals K, de Keyser F, van derStraeten C, Verbruggen G, et al. Onset of symptoms of rheuma-toid arthritis in relation to age, sex and menopausal transition.J Rheumatol 1990;17:1620–2.

36. Holmdahl R, Jansson L, Andersson M. Female sex hormonessuppress development of collagen-induced arthritis in mice. Ar-thritis Rheum 1986;29:1501–9.

37. Cevik R, Em S, Gur A, Nas K, Sarac AJ, Colpan L. Sex and thyroidhormone status in women with rheumatoid arthritis: are there anyeffects of menopausal state and disease activity on these hor-mones? Int J Clin Pract 2004;58:327–32.

38. Robinzon B, Cutolo M. Should dehydroepiandrosterone replace-ment therapy be provided with glucocorticoids? Rheumatol (Ox-ford) 1999;38:488–95.

39. Ostensen M, Villiger PM. Immunology of pregnancy: pregnancy asa remission inducing agent in rheumatoid arthritis. Transpl Immu-nol 2002;9:155–60.

40. Castagnetta LA, Carruba G, Granata OM, Stefano R, Miele M,Schmidt M, et al. Increased estrogen formation and estrogen toandrogen ratio in the synovial fluid of patients with rheumatoidarthritis. J Rheumatol 2003;30:2597–605.

41. Scorza R, Caronni M, Bazzi S, Nador F, Beretta L, Antonioli R, etal. Post-menopause is the main risk factor for developing isolated

pulmonary hypertension in systemic sclerosis. Ann N Y Acad Sci2002;966:238–46.

42. Suzuki T, Schaumberg DA, Sullivan BD, Liu M, Richards SM,Sullivan RM, et al. Do estrogen and progesterone play a role in thedry eye of Sjogren’s syndrome? Ann N Y Acad Sci 2002;966:223–5.

43. Toloza SM, Uribe AG, McGwin G Jr, Alarcon GS, Fessler BJ,Bastian HM, et al, for the LUMINA Study Group. Systemic lupuserythematosus in a multiethnic US cohort (LUMINA). XXIII.Baseline predictors of vascular events. Arthritis Rheum 2004;50:3947–57.

44. Sander GE, Giles TD. Cardiovascular complications of collagenvascular disease. Curr Treat Options Cardiovasc Med 2002;4:151–9.

45. Rossouw JE, Anderson GL, Prentice RL, Lacroix AZ, KooperbergC, Stefanick ML, et al. Risks and benefits of estrogen plusprogestin in healthy postmenopausal women: principal resultsfrom the Women’s Health Initiative randomized controlled trial.J Am Med Assoc 2002;288:321–33.

46. Manson JE, Hsia J, Johnson KC, Rossouw JE, Assaf AR, LasserNL, et al. Women’s Health Initiative investigators: estrogen plusprogestin and the risk of coronary heart disease. N Engl J Med2003;349:523–34.

47. Grady D, Herrington D, Bittner V, Blumenthal R, Davidson M,Hlatky M, et al. Cardiovascular disease outcomes during 6.8 yearsof hormone therapy: heart and estrogen/progestin replacementstudy follow-up (HERS II). J Am Med Assoc 2002;288:49–57.

48. Katz RJ, Hsia J, Walker P, Jacobs H, Kessler C. Effects ofhormone replacement therapy on the circadian pattern of athero-thrombotic risk factors. Am J Cardiol 1996;78:876–80.

49. Salpeter S, Walsh JM, Greyber E, Ormiston TM, Salpeter EE.Mortality associated with hormone replacement therapy inyounger and older women: a meta-analysis. J Gen Intern Med2004;19:791–804.

50. Stencheber M, Droegemueller W, Herbst A, Misnell D. Meno-pause. In: Stencheber M, editor. Comprehensive gynecology. St.Louis: Mosby; 2000. p. 1217–50.

51. Gladman D, Ginzler E, Goldsmith C, Fortin P, Liang M, UrowitzM, et al. The development and initial validation of the SystemicLupus International Collaborating Clinics/American College ofRheumatology damage index for systemic lupus erythematosus.Arthritis Rheum 1996;39:363–9.


Systemic lupus erythematosus in a multiethnic US cohort (LUMINA): XXI. Disease activity, damage...

Documents

Transcript of Systemic lupus erythematosus in a multiethnic US cohort (LUMINA): XXI. Disease activity, damage...