1. Introduction
Systemic lupus erythematosus (SLE) is a typical autoimmune connective tissue disease that can clinically accumulate in multiple organs throughout the body and is more prevalent in women of reproductive age [
1,
2]. Sjögren’s syndrome (SS) is a chronic inflammatory autoimmune disease mainly involving the exocrine glands. SS is clinically manifested as dryness of the mouth and eyes, but extra glandular organs may also be involved, either independently or in combination with other connective tissue diseases, including rheumatoid arthritis and systemic lupus erythematosus [
3,
4,
5]. These two diseases often coexist in the same individual, which may affect the proper diagnosis and prognosis for these patients. The association of SLE and SS was first described in 1959, and SLE with SS is considered a distinct subtype of SLE [
6]. In recent years, the number of patients with SLE and SS overlap has gradually increased, and studies have shown that 8.3–19.0% of SLE patients also have SS [
7]. Another study in China even showed that 24 out of 55 SLE cases (43.6%) had secondary SS [
8].
In addition to finding that the prognosis of patients with SLE-SS overlap is different from that of patients with a single disease, previous studies have also revealed that these patients have unique clinical characteristics [
9,
10,
11]. Unfortunately, the conclusions in different reports are varied, and more than 10 types of clinical manifestations are involved, which is difficult to use for routine clinical implementation. Therefore, it is necessary to develop a practical and reliable screening technique to help identify patients with SLE-SS overlap early to improve their long-term outcome.
In this study, we retrospectively analyzed clinical data related to patients with SLE, SS, and SLE-SS. The risk factors affecting patients with SLE or SS in SLE-SS were extracted, and a decision model for predicting patients with SLE or SS versus SLE-SS was constructed. This model may provide a simple and effective tool for identifying patients with SLE-SS.
2. Materials and Methods
2.1. Patients and Controls
Data were retrospectively collected from 477 patients with SLE (
n = 211), pSS (
n= 187), and SLE-SS (
n = 79) between December 2018 and December 2020 from the Department of Rheumatology and Immunology of Nanjing Drum Tower Hospital. The diagnosis of SLE was given according to the 1997 American College of Rheumatology (ACR) revised criteria for the classification of systemic lupus erythematosus [
12], and the diagnosis of pSS was based on the 2002 American-European Consensus group criteria [
13]. SLE-SS is defined as meeting the diagnostic criteria for both SLE and secondary SS, excluding patients with lymphoma, nodal disease, hepatitis virus infection, AIDS, radiation therapy, and anti-acetylcholine drugs.
2.2. Data Collection
Data on demographic characteristics and clinical and laboratory findings were collected retrospectively. The clinical features assessed included fever, alopecia, asthenia, dry mouth, dry eye, photosensitivity, skin lesions, oral ulcer, Raynaud phenomenon, cavities, epistaxis, arthritis, interstitial lung disease, pulmonary arterial hypertension, and vasculitis. Laboratory findings included proteinuria, hematuria, anemia, leukocytopenia, thrombocytopenia, hypoalbuminemia, increased blood urea nitrogen, increased serum creatinine, increased erythrocyte sedimentation rate (ESR), increased C-reactive protein, hypocomplementemia, ANA, anti-dsDNA antibody, anti-SSA antibody, anti-Ro52 antibody, anti-SSB antibody, anti-Sm antibody, anti-RNP antibody, anti-cardiolipin antibody, rheumatoid factor (RF)-positive, and IgG elevation. Normal values for the laboratory tests were as follows: hemoglobin ≥ 110 g/L (female) or 120 g/L (male); platelets 100–300 × 109/L; leukocytes 4–10 × 109/L; serum albumin ≥ 35 g/L; blood urea nitrogen (BUN) ≤ 7.5 mmol/L; serum creatinine ≤ 133 μmol/L; erythrocyte sedimentation rate (ESR) ≤ 20 (female) or ≤15 mm/h (male); CRP ≤ 8 mg/L; complement C3 ≥ 0.8 g/L or C4 ≥ 0.2 g/L; ANA-negative (<1:40); anti-dsDNA-negative, anti-SSA-negative, anti-Ro52-negative, anti-SSB-negative, anti-Sm-negative, and anti-RNP-negative; anti-cardiolipin antibody < 12 U/mL or negative; RF < 20 IU/mL; IgG ≤ 16 g/L; and urine protein < 0.5 g/24 h or less than 2+. All the antibodies tested were IgG type, and positivity and negativity were defined according to the standard in the Drum Tower Hospital.
2.3. Statistical Analysis
Categorical variables are presented as numbers and standard frequencies and were tested by the χ2 test or Fisher’s exact test. Continuous variables are presented as medians and interquartile ranges (IQRs) and were compared using the Mann–Whitney U test.
For SLE versus SLE-SS, a total of 290 patients were randomly assigned to the discovery cohort (
n = 203) or the validation cohort (
n = 87) using a computer program at a ratio of 7:3. Similarly, for SS versus SLE-SS, a total of 266 patients were randomly assigned to the discovery cohort (
n = 187) or the validation cohort (
n = 79) using a computer program at a ratio of 7:3. For the training group, the sample size followed the 10-fold criterion, that is, each prediction variable needs at least 10 observations to produce a reasonable and stable estimate [
14,
15]. In this study, 5 predictive variables were selected as the final model, requiring at least 50 observations and a sample size of at least 200 for the training group based on a 25% incidence. PASS 15 software (NCSS, LLC, Kaysville, Utah) was used to calculate the sample size of the verification group [
16]. The minimum number of positive events was 10, the minimum number of negative events was 50, and the minimum sample size of the verification group was 60. A sufficient sample size is included in this study. We developed prediction tools using the discovery cohorts. A least absolute shrinkage and selection operator (LASSO) logistic regression model and 10-fold cross-validation were applied to select the best predictive features among the clinical characteristics and laboratory parameters in the discovery cohort that were statistically different between SLE and SLE-SS or SS and SLE-SS. The performance of the predictive tool was evaluated for discrimination, calibration, and clinical usefulness. To assess the discrimination of the predictive model, the discriminatory ability was measured using the area under the receiver operating characteristic (AUC) curve. Calibration curves were then plotted to evaluate the calibration effect of the SLE-SS prediction models. Decision curve analysis was used by quantifying the net benefits to determine the clinical utility of the SLE-SS predictive models.
All data analyses used R software (version 4.0.3). In R software, Packages (“rms”) and (“rmda”) were operated. All tests were two-sided, and a p-value < 0.05 was considered statistically significant.
3. Results
3.1. Patient Characteristics
For SLE versus SLE-SS, a total of 290 patients were included in this study, and a random sample at a ratio of 7:3 was divided into discovery (
n = 203) and validation groups (
n = 87). Regarding SS versus SLE-SS, a total of 266 patients were included in this study and were randomized in a 7:3 ratio to the discovery (
n = 187) and validation groups (
n = 79). A flow chart is shown in
Figure 1.
Table S1 summarizes the clinical characteristics and laboratory parameters of the discovery and validation groups. The characteristics of the two groups of patients were comparable. We also summarize the clinical characteristics and laboratory parameters of SLE, SS, and SLE-SS in the discovery and validation cohorts in
Tables S2 and S3.
3.2. Development of the Prediction Tool for SLE or SLE-SS
The 11 candidate clinical and laboratory indicators that were statistically different between SLE-SS and SLE were reduced to the five most useful predictive markers by using LASSO logistic regression (
Figure 2A,B). These characteristics included dry mouth, dry eye, anti-Ro52 positive, anti-SSB positive, and RF positive. The results of the logistic regression analysis for SLE versus SLE-SS are given in
Table 1. We constructed prediction tools based on these factors for SLE in SLE-SS and visualized them as nomograms (
Figure 3A).
Similarly, as shown in
Figure 2C,D, LASSO logistic regression was used to reduce the 17 candidate laboratory indicators that were statistically different between SLE-SS and SS to the 11 most useful predictive markers including age at diagnosis (years), fever, dry mouth, photosensitivity, skin lesions, arthritis, proteinuria, hematuria, hypoalbuminemia, anti-dsDNA positive, and anti-Sm positive. The results of the logistic regression analysis of SS and SLE-SS are shown in
Table 1. Based on these factors, we constructed a prediction tool for SS in SLE-SS and visualized it as a nomogram (
Figure 3B).
3.3. Model Performance Assessment
The AUC of SLE versus SLE-SS and SS versus SLE-SS was 0.880 (95% CI: 0.826–0.934) and 0.970 (95% CI: 0.946–0.995), respectively, indicating that the predictive tool had good discrimination in the discovery cohort (
Figure 4A,D). In this study, the calibration curves for both SLE versus SLE-SS and SS versus SLE-SS also showed good nomogram agreement in the discovery cohorts (
Figure 4C,F). The DCA of the SLE versus SLE-SS prediction tool is presented in
Figure 5A. The decision curve showed that if the threshold probability of a patient and a doctor was >5 and <78%, respectively, the use of this SLE versus SLE-SS nomogram to predict the risk of SLE-SS added more benefit than the scheme. The DCA of the SS versus SLE-SS prediction tool indicated that the prediction tool conferred more net benefits than the SLE-SS patients’ scheme across all threshold probabilities (
Figure 5C). The validation cohort provided reconfirmation that the prediction tool had good discrimination (AUC: 0.865; 95% CI: 0.781–0.949,
Figure 4B; AUC: 0.936; 95% CI: 0.888–0.984,
Figure 4E), good calibration (
Figure 4C,F), and clinical utility (
Figure 5B,D).
4. Discussion
Previous studies have explored potential risk factors for SLE or SS in SLE-SS, including demographics, clinical characteristics, and laboratory parameters [
10,
11,
17,
18]. However, no studies have investigated models that incorporate multiple laboratory findings in clinical decision making. In this study, the SLE-SS overlapping disease prediction models were established using the minimum absolute contraction and selection operator (LASSO) logistic regression model between the candidate clinical features and laboratory parameters and were visualized as nomograms to help predict the risk of SLE or SS in SLE-SS.
The clinical features of SS versus SLE-SS, such as dry mouth, dry eye, and a unique autoantibody profile associated with pSS, were identified. In Yang’s study, we found that the frequency of dry mouth and dry eye was significantly higher in SLE-SS than in the SLE group [
18]. Most previous studies have shown that anti-SSA/Ro and anti-SSB/La antibodies are significantly more positive in SLE-SS than in the SLE group [
19]. Regarding RF positivity, Santos and Manoussakis et al. showed that RF positivity was significantly higher in both SLE-SS groups than in the SLE group [
20]. Our results were largely consistent with previous studies. We conclude that dry mouth, dry eye, anti-Ro52 positive, anti-SSB positivity, and RF positivity are risk factors for SLE progressing to SLE-SS. Therefore, we should be aware of the development of combined SS when young SLE patients present with dry mouth, dry eye, or positive anti-Ro52, anti-SSB, or RF.
Meanwhile, patients with SLE-SS also have some clinical characteristics and autoantibody profiles specific to SLE. Szanto and Manoussakis et al. showed a higher rate of anti-dsDNA antibody positivity in SLE-SS than in the SS group and a significantly higher frequency of hypoalbuminemia in SLE-SS than in the SS group [
10,
11]. Patients with SLE-SS are also more likely to have renal involvement, such as proteinuria and hematuria. The study by Szanto and Yang et al. suggests that SLE-SS patients are more likely to develop renal damage than patients with SS [
20,
21]. In our study, we constructed a predictive model with risk factors such as proteinuria, hematuria, hypoalbuminemia, and positive anti-dsDNA antibodies, suggesting that when SS patients present with these symptoms, they are more likely to have combined SLE.
There are several limitations of this study. First, our collected data are retrospective, and various biases are inevitable due to the nature of the study itself, even though all three groups of patients were subjected to the same conditions and criteria to avoid selection bias and confounding factors. Second, the sample size was relatively small, and we welcome external validation in a wider population.
5. Conclusions
Based on the patient’s clinical symptoms, the development of SS should be monitored when SLE patients are positive for anti-RO52, anti-SSB, and RF. However, the presence of SS patients with albuminuria, hematuria, hypoproteinemia, positive anti-dsDNA, and positive anti-SM may indicate the presence of SLE. On the basis of clinical symptoms and serum indicators, this study established a risk prediction model for SLE or SS transitioning into SLE-SS with high accuracy, which may help with early diagnosis and the selection of appropriate treatment.
Supplementary Materials
The following supporting information can be downloaded at:
https://www.mdpi.com/article/10.3390/jcm12020535/s1, Table S1: Patient demographic and clinical characteristics in the discovery and validation cohorts; Table S2: Clinical characteristics of SLE and SLE-SS in the discovery and validation cohorts; Table S3: Clinical characteristics of SS and SLE-SS in the discovery and validation cohorts.
Author Contributions
Y.H. and L.M. collected the data. Y.H. analyzed the results and drafted the manuscript. X.F. designed and supervised the project and edited the manuscript. Z.J., D.W., Y.Z., B.H., H.W. and S.C. helped with data collection and analysis. All authors have read and agreed to the published version of the manuscript.
Funding
This work was supported by the National Key Research and Development Program of China (2019YFE0111700) and the National Natural Science Foundation of China (No.81971517).
Institutional Review Board Statement
The study was approved by the Ethics Committee of Drum Tower Hospital (No. 2020-093).
Informed Consent Statement
The consent is waived because it is a retrospective chart review study.
Data Availability Statement
Data are available on request from the corresponding author.
Conflicts of Interest
The authors declare no conflict of interest.
References
- Kiriakidou, M.; Ching, C.L. Systemic Lupus Erythematosus. Ann. Intern. Med. 2020, 172, 81–96. [Google Scholar] [CrossRef] [PubMed]
- Ramaswamy, M.; Tummala, R.; Streicher, K.; da Costa, A.N.; Brohawn, P.Z. The Pathogenesis, Molecular Mechanisms, and Therapeutic Potential of the Interferon Pathway in Systemic Lupus Erythematosus and Other Autoimmune Diseases. Int. J. Mol. Sci. 2021, 22, 11286. [Google Scholar] [CrossRef] [PubMed]
- Stefanski, A.-L.; Tomiak, C.; Pleyer, U.; Dietrich, T.; Burmester, G.R.; Dörner, T. The Diagnosis and Treatment of Sjögren’s Syndrome. Dtsch. Arztebl. Int. 2017, 114, 354–361. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Fox, R.I. Sjögren’s syndrome. Lancet 2005, 366, 321–331. [Google Scholar] [CrossRef]
- Psianou, K.; Panagoulias, I.; Papanastasiou, A.D.; de Lastic, A.-L.; Rodi, M.; Spantidea, P.I.; Degn, S.E.; Georgiou, P.; Mouzaki, A. Clinical and immunological parameters of Sjögren’s syndrome. Autoimmun. Rev. 2018, 17, 1053–1064. [Google Scholar] [CrossRef]
- Heaton, J.M. Sjögren’s syndrome and systemic lupus erythematosus. Br. Med. J. 1959, 1, 466–469. [Google Scholar] [CrossRef]
- Gilboe, I.-M.; Kvien, T.K.; Uhlig, T.; Husby, G. Sicca symptoms and secondary Sjögren’s syndrome in systemic lupus erythematosus: Comparison with rheumatoid arthritis and correlation with disease variables. Ann. Rheum. Dis. 2001, 60, 1103–1109. [Google Scholar] [CrossRef] [Green Version]
- Lei, Y.; Weihong, Y.; Gangbo, L.; Jinqi, Z. Salivary gland TcmO_4 in patients with systemic lupus erythematosus combined with Sjögren’s syndrome imaging study. J. Kunming Med. Coll. 2007, 28, 59–61. [Google Scholar]
- Hernández-Molina, G.; Zamora-Legoff, T.; Romero-Díaz, J.; Nuñez-Alvarez, C.A.; Cárdenas-Velázquez, F.; Hernández-Hernández, C.; Calderillo, M.L.; Marroquín, M.; Recillas-Gispert, C.; Ávila-Casado, C.; et al. Predicting Sjögren’s syndrome in patients with recent-onset SLE. Rheumatology 2013, 52, 1438–1442. [Google Scholar] [CrossRef] [Green Version]
- Santos, C.S.; Morales, C.M.; Castro, C.; Alvarez, E.D. Polyautoimmunity in systemic lupus erythematosus: Secondary Sjogren syndrome. Z. Rheumatol. 2021. [Google Scholar] [CrossRef]
- Manoussakis, M.N.; Georgopoulou, C.; Zintzaras, E.; Spyropoulou, M.; Stavropoulou, A.; Skopouli, F.N.; Moutsopoulos, H.M. Sjögren’s syndrome associated with systemic lupus erythematosus: Clinical and laboratory profiles and comparison with primary Sjögren’s syndrome. Arthritis Rheum. 2004, 50, 882–891. [Google Scholar] [CrossRef] [PubMed]
- Hochberg, M.C. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum. 1997, 40, 1725. [Google Scholar] [CrossRef]
- Vitali, C.; Bombardieri, S.; Jonsson, R.; Moutsopoulos, H.M.; Alexander, E.L.; Carsons, S.E.; Daniels, T.E.; Fox, P.C.; Fox, R.I.; Kassan, S.S.; et al. Classification criteria for Sjögren’s syndrome: A revised version of the European criteria proposed by the American-European Consensus Group. Ann. Rheum. Dis. 2002, 61, 554–558. [Google Scholar] [CrossRef] [Green Version]
- Van Smeden, M.; Moons, K.G.; De Groot, J.A.; Collins, G.S.; Altman, D.G.; Eijkemans, M.J.; Reitsma, J.B. Sample size for binary logistic prediction models: Beyond events per variable criteria. Stat. Methods Med. Res. 2019, 28, 2455–2474. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Hillberg, N.S.; Hogenboom, J.; Hommes, J.; Van Kuijk, S.; Keuter, X.; van der Hulst, R. Risk of major postoperative com-plications in breast reconstructive surgery with and without an acellular dermal matrix: A development of a prognostic pre-diction model. JPRAS Open 2022, 33, 92–105. [Google Scholar] [CrossRef] [PubMed]
- Feng, D.; Cortese, G.; Baumgartner, R. A comparison of confidence/credible interval methods for the area under the ROC curve for continuous diagnostic tests with small sample size. Stat. Methods Med. Res. 2017, 26, 2603–2621. [Google Scholar] [CrossRef]
- Ruacho, G.; Kvarnström, M.; Zickert, A.; Oke, V.; Rönnelid, J.; Eketjäll, S.; Elvin, K.; Gunnarsson, I.; Svenungsson, E. Sjögren Syndrome in Systemic Lupus Erythematosus: A Subset Characterized by a Systemic Inflammatory State. J. Rheumatol. 2020, 47, 865–875. [Google Scholar] [CrossRef]
- Yao, Q.; Altman, R.D.; Wang, X. Systemic lupus erythematosus with Sjögren syndrome compared to systemic lupus erythematosus alone: A meta-analysis. J. Clin. Rheumatol. 2012, 18, 28–32. [Google Scholar] [CrossRef]
- Baer, A.N.; Maynard, J.W.; Shaikh, F.; Magder, L.S.; Petri, M. Secondary Sjogren’s syndrome in systemic lupus erythematosus defines a distinct disease subset. J. Rheumatol. 2010, 37, 1143–1149. [Google Scholar] [CrossRef]
- Szanto, A.; Szodoray, P.; Kiss, E.; Kapitany, A.; Szegedi, G.; Zeher, M. Clinical, serologic, and genetic profiles of patients with associated Sjögren’s syndrome and systemic lupus erythematosus. Hum. Immunol. 2006, 67, 924–930. [Google Scholar] [CrossRef]
- Yang, Y.; Li, Z.; Wang, L.; Zhang, F. The clinical and laboratory characteristics of Sjögren’s syndrome that progresses to systemic lupus erythematosus: A retrospective case-control study. Int. J. Rheum. Dis. 2013, 16, 173–177. [Google Scholar] [CrossRef] [PubMed]
| Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).