Associates and predictors of pleurisy or pericarditis in SLE

Background/Purpose Serositis is one of both ACR and SLICC classification criteria for systemic lupus erythematosus (SLE) and a common type of extra renal flare. However, little is known about clinical or immunological associations of pleurisy or pericarditis. The aim of this study is to analyze associates and predictors of pleurisy versus pericarditis in Caucasians and African Americans with SLE. Methods 2,390 SLE patients in the Hopkins Lupus Cohort were analyzed for demographic, clinical and serologic associates of pleurisy or pericarditis, defined using the SELENA revision of the SLE Disease Activity Index (SLEDAI). The cross-sectional and prospective study using either univariate or multivariate analysis were performed to evaluate the associates of serositis in SLE. We reported associates with a p-value of less than 0.05 for pleurisy or pericarditis. Results 43% had pleurisy and 22% had pericarditis. African-American ethnicity was a predictive factor for new pericarditis. Hemolytic anemia, proteinuria, lymphadenopathy and anti-Sm were predictive only of pericarditis, whereas pulmonary fibrosis and GI infarction were predictive only of pleurisy. Fever, Raynaud’s syndrome, and anti-DNA were predictors for both pericarditis and pleurisy. Conclusion Our study provides further insights into the associates of pleurisy and pericarditis in SLE. Predictors of pleurisy and pericarditis are shown for the first time. The long term consequences from the cross-sectional analysis gives a lesson that serositis in SLE should not be considered benign.

The Hopkins Lupus Cohort allowed us to study serositis in a large number of patients with SLE. We examined associations of EVER having pleurisy or pericarditis and of developing NEW pleurisy or pericarditis after cohort entry. In particular, we were able to examine associated or predictive factors in both Caucasian and African-American patients, and to find separate associated or predictive factors of pleurisy and pericarditis.

Methods study group and diagnostic criteria
There were 2390 patients with SLE enrolled in the Hopkins Lupus Cohort at the time of the analysis. The cohort was approved by the Johns Hopkins University School of Medicine Institutional Review Board on a yearly basis. All participants signed informed consent. 16 We characterised patients with pericarditis or pleurisy, according to the definition of the Safety of Estrogens in Lupus Erythematosus National Assessment-SLE Disease Activity Index (SELENA-SLEDAI). 17 Pericarditis was diagnosed when the patients had pericardial pain, pericardial friction rub or Lupus Science & Medicine pericardial effusion or ECG or echocardiogram confirmation of pericarditis. Pleurisy was diagnosed by pleuritic chest pain, pleural rub or pleural effusion. As we used the SLEDAI, we were not able to do separate analysis of pleural rub or effusion.

Associates and predictors of pleurisy or pericarditis in SLE
At cohort entry, basic demographic characteristics (age, age at cohort entry, age at SLE diagnosis, years of education, sex, ethnicity and smoking history), clinical features and serologic data (including lupus anticoagulant, anticardiolipin, Coombs test, anti-dsDNA, anti-La, anti-RNP, anti-Sm, complement C3 and C4 levels and erythrocyte sedimentation rate (ESR)) were collected, and updated at each follow-up visit (at least quarterly).

statistical analysis
Patients with serositis were compared with patients without serositis, with respect to demographic, clinical and serologic characteristics. For continuous variables (age, age at cohort entry, age at SLE diagnosis, years of education and SDI score), the Student's t-test was used to compare means. Pearson's Χ 2 test was used to analyse categorical variables. Fisher's exact test was used, instead of Pearson's, when >20% of the cells in 2×2 contingency table had a frequency <5. The significance level (α) was 0.05 for two-tailed tests. The ORs and 95% CIs were computed to compare the strength of the association.
Cox proportional hazard models were applied to evaluate the association (HR) between demographic characteristics, clinical and serologic features, and newly diagnosed serositis during follow-up. Univariate Cox proportional hazard models were done with adjustment for age and ethnicity. Bonferroni correction was used to account for multiple comparisons in the univariate Of the 2390 patients with SLE in the cohort, 2132 patients (89%) were included in the prospective analysis, after excluding those who had serositis before SLE diagnosis. Among the prospective subset, 421 (20%) had pericarditis and the median follow-up time (including censored data) was 106 months (IQR 39. 4-191.5). Seven hundred thirty-seven (35%) had new pleurisy and the median follow-up time (including censored data) was 108 months (30.0-167.3).
Associates from cross-sectional study and predictors from prospective study of demographic, clinical, serologic and organ damage-related variables for pericarditis or pleurisy are summarised in tables 2 and 3. Except otherwise noted, only significantly (p<0.0001) associated factors with either pericarditis or pleurisy were mentioned in this section.
Cross-sectional study: association of demographic characteristics with serositis in sle African-American ethnicity was highly associated only with pericarditis, and age at SLE diagnosis was significantly associated only with pleurisy. There was no demographic factor that was significantly associated with both pericarditis and pleurisy.
Private health insurance (OR 0.75, p=0.0151) was the only protective factor against pericarditis among demographic characteristics. Gender, age at cohort entry and smoking history were not associated with either pleurisy or pericarditis.
Cross-sectional study: association of clinical features with serositis in sle Factors significantly associated only with pericarditis were nephrotic syndrome, proteinuria, seizure and haematuria. Arthralgia, pancreatitis, arthritis and malar rash were only significantly associated with pleurisy. Factors highly associated with both pleurisy and pericarditis were pulmonary hypertension, fever, pulmonary fibrosis, haemolytic anaemia, lymphadenopathy, Raynaud's syndrome and anaemia.
Cross-sectional study: association of serologic features with serositis in sle Only pericarditis was highly associated with anti-Sm and anti-RNP. Serologic tests significantly associated with both pericarditis and pleurisy were high ESR, anti-DNA, low C3 and low C4.
Cross-sectional study: association of organ damage with serositis in sle Organ damage that was significantly associated with only pericarditis included renal insufficiency, pulmonary hypertension, pleural fibrosis and cardiomyopathy. Deforming or erosive arthritis was associated only with pleurisy. Pulmonary fibrosis, and gastrointestinal infarction or resection were highly associated with both pleurisy and pericarditis.
Prospective study: predictive factors for serositis in patients with sle A prospective analysis using univariate and multivariate Cox proportional hazard model was done to investigate predictive factors for new serositis in patients with SLE.
The univariate Cox regression model was adjusted for age and ethnicity. The results are shown in the right column of the tables 2 and 3. The analysis of either age or ethnicity for the univariate model was done independently, so as not to miss their own predictive value. For pericarditis, compared with Caucasians, the HR for African-Americans was 1.91 (95% CI 1.56 to 2.33, p<0.0001) and the HR for 1 year increase in age is 0.99 (95% CI 0.98 to 1.00, p=0.0033). For pleurisy, compared with Caucasians, the HR for African-Americans was 1.32 (95% CI 1.13 to 1.53, p=0.0003) and the HR for 1 year increase was 0.99 (95% CI 0.98 to 0.99, p=0.0002).
The results of the multivariate Cox regression (p<0.05) using only covariates from the univariate analysis are shown in tables 4 and 5. African-American, haemolytic anaemia, proteinuria, lymphadenopathy and anti-Sm were predictive factors only for pericarditis. Arthritis, pulmonary fibrosis, low C4 and infarction or resection of bowel were predictors of only pleurisy. Predictors of both pericarditis and pleurisy were fever, Raynaud's syndrome and anti-DNA.

dIsCussIon
Although cross-sectional analyses of serositis have been done in SLE, ours is the first prospective analysis of new serositis and the first study to emphasise the different associates and predictors of pericarditis versus pleurisy.
It is important to compare the results between the cross-sectional analysis and prospective analyses. The cross-sectional analysis includes all patients, and thus has Lupus Science & Medicine  20 reported that 'primary cardiac disease', including pericarditis (81.2% of all cases), was associated with African-Latin American ethnicity. Because serositis had been thought to be one pathogenic mechanism, it is quite surprising (and not understood) why pericarditis is associated with African-American ethnicity, whereas pleurisy is not. Gender was not associated with either pericarditis or pleurisy in either cross-sectional or prospective analysis.
In contrast, a Chinese cohort 11 found an association with male gender, but only in the age group of 15-50 years. We did not have enough Asian patients to do a subset analysis.
Our cross-sectional analysis showed that younger age at SLE diagnosis was associated with both pericarditis and pleurisy. A Canadian, multiethnic study 15 also found pleurisy to be associated with younger age at SLE diagnosis. A Chinese cohort, 11 however, found no association of age at SLE onset with either pericarditis or pleurisy. It should be noted that younger age at SLE diagnosis was not predictive of either new pericarditis and pleurisy in our prospective analysis.
Smoking history was not associated with either pericarditis or pleurisy, consistent with one previous report. 15 However, one study 21 of 119 patients did find an association with current smoking. The study by Rubin et al, 21 however, only included patients <1 year after diagnosis, and female patients with no history of therapy.
We found that only few variables were shown in the result of multiple variable analysis (tables 4 and 5): age at SLE diagnosis (pericarditis and pleurisy: cross-sectional study) and African-American (pericarditis: cross-sectional and prospective study).

Epidemiology and outcomes
Clinical associates/predictors of serositis in sle We found that haemolytic anaemia was associated with both pericarditis and pleurisy, consistent with the previous result of a multiethnic group study 12 of 1251 patients. Haemolytic anaemia was predictive of only pericarditis in the multivariate prospective analysis. According to our univariate cross-sectional analysis, livedo reticularis was a protective factor only for pericarditis (OR=0.74, p=0.0108). Livedo reticularis is commonly found in patients with SLE with antiphospholipid antibodies. 22 However, in our study, lupus anticoagulant was not protective against pericarditis. Anticardiolipin was also not protective against pericarditis or pleurisy, similar to a Chinese study. 14 Therefore, it is highly unlikely that the protective association with livedo reticularis was explained by a confounding effect of antiphospholipid antibodies. The protective mechanism of livedo reticularis for pericarditis is not understood.
Central nervous system (CNS) involvement was reported to be a protective factor (OR=0.52, p=0.0302) for pulmonary manifestations, including pleural effusion, in a Latin American study. 20 We, however, could not find any prospective association of CNS involvement with pericarditis or pleurisy in any of our analysis.
Only three clinical factors (proteinuria, lymphadenopathy and Raynaud's syndrome) among variables in tables 2 and 3 were associates and predictors for both pericarditis and pleurisy in the multiple variable analysis (tables 4 and 5).

serologic associates/predictors of serositis in sle
The association of anti-DNA with both pericarditis and pleurisy in the univariate cross-sectional and prospective analyses in our study was in contradiction to a previous Chinese cohort study of 917 patients. 14 Pleural effusion was shown to be associated with anti-DNA in one previous study from Saudi Arabia of 61 patients. 23 Low complement (C3 and C4) was associated with both pericarditis and pleurisy in our univariate cross-sectional and prospective analyses. Other investigators 24 observed a relatively weak association of pericardial tamponade with low C4 (p=0.05), compared with pericardial effusions as a whole. In terms of pleurisy, in a Saudi Arabian cohort of 61 patients, 23 pleural effusion was associated with both low C3 and C4.

Lupus Science & Medicine
ESR was associated with both pericarditis and pleurisy in the univariate cross-sectional analysis. Previously, using the Hopkins Lupus Cohort, we 25 reported that ESR was correlated with serositis. We now can extend this previous result to an association with both pericarditis and pleurisy individually.
Anti-RNP was associated with both pericarditis and pleurisy in cross-sectional univariate analyses, but not the prospective analysis. An association with pleurisy was shown in the Canadian cohort, 15 but not in a Chinese cohort of 917 patients. 14 In our univariate analysis, anti-Sm was associates and predictor of pericarditis and pleurisy, in contrast to a previous  Epidemiology and outcomes report. 14 Anti-Sm was a predictive factor for only pericarditis in the multiple variable model, as well. The association of anti-Sm with pleurisy was, however, supported by the study of the Canadian cohort. 15 Our multivariable analysis (tables 4 and 5) for serologic variables showed that anti-DNA and anti-Sm were both associates and predictors for pericarditis. They were factors shown in only one study for pleurisy: anti-DNA (prospective), low C4 (prospective) and low C3 (cross-sectional study).
damage associates/predictors of serositis in sle Permanent organ damage was assessed by SDI. The multivariable, prospective study showed that pleurisy predicted later gastrointestinal infarction or resection, but there was no factor predictive of pericarditis.

Conclusion
Our study shows, for the first time, the different predictors of pleurisy and pericarditis in SLE. The cross-sectional analysis also defines the long-term consequences, especially of pulmonary fibrosis and infarction of bowel. Serositis in SLE should not be considered benign, as it was associated with long-term organ damage in SLE.

strengths and limitations
The strengths of our study are the large number of patients, the ethnic diversity and the cohort structure with quarterly follow-up and protocolised laboratory assessment. Predictors from the prospective analysis are anticipated to give the most important information for clinicians. We are the first to separately analyse associates and predictors of pleurisy versus pericarditis.
This study still has several limitations. We did not use cardiac ultrasound or chest X-ray when the clinical diagnosis of serositis was secure. Our results may not be generalisable to the most severe forms of serositis in patients with SLE. To avoid the issue of multiple comparisons, we have emphasised variables that met the most stringent p value cut-off, and thus may have missed some borderline associations.
Contributors MP acquired the data in the cohort. SR, WF and MP designed the study, analysed and interpreted the data and revised the manuscript. All authors approved the final version of the manuscript.
Funding The Hopkins Lupus Cohort was funded by NIH AR043727 and AR069572. open Access This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http:// creativecommons. org/ licenses/ by/ 4. 0/