{"paper_id":"d57e2314-70fa-4f3e-ba4c-7e0573257219","body_text":"1\nVol.:(0123456789)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports\nAssociation between endometriosis \nand risk of systemic lupus \nerythematosus\nYu‑Hsi Fan1,8, Pui‑Ying Leong2,3,8, Jeng‑Yuan Chiou4, Yu‑Hsun Wang5, Ming‑Hsiang Ku6 & \nJames Cheng‑Chung Wei2,3,7*\nTo examine the association between endometriosis and the risk of systemic lupus erythematosus \n(SLE), this nationwide, population‑based, retrospective cohort study was conducted based on \nNational Health Insurance Research Database in Taiwan. Endometriosis (N = 16,758) and non‑\nendometriosis (N = 16,758) groups were identified by matching baseline characteristics and \ncomorbidities. Student’s t‑tests and the Kaplan–Meier estimator were utilized to estimate the hazard \nratio (HR) and cumulative probability of SLE in the two groups. The endometriosis group showed \na significantly higher incidence density rate (0.3 vs. 0.1 per 1000 person‑years) and hazard ratio in \nSLE group (adjusted HR [aHR], 2.37; 95% confidence interval [CI] 1.35–4.14) compared to the non‑\nendometriosis group. Subgroup analysis revealed that patients with endometriosis between 30 and \n45 years of age, or were non‑steroidal anti‑inflammatory drug users, or were hormonal medications‑\nfree participants, had higher risks of SLE. For patients with endometriosis, surgical intervention \ndid not significantly impact on the risk of SLE. Our results demonstrated an increased risk of SLE in \npatients with endometriosis. Clinicians should be aware of this association when managing patients \nwith endometriosis or SLE.\nEndometriosis is one of the most common gynaecological disorders that affects nearly 10% of menstruating \n women1. Pelvic pain and infertility are common symptoms of  endometriosis2. Although the causes of endome-\ntriosis remain uncertain, one hypothesis is retrograde menstruation. It is assumed that, as a result of retrograde \nmenstruation, endometrial tissues adhere to pelvic organs and peritoneal surfaces, which subsequently become \nectopic endometrial tissues. Ectopic endometrial tissues, just like the endometrium, proliferate and shed as they \nare regulated by the level of oestrogen and progesterone. The exfoliating tissues stimulate the inflammatory \nresponse and result in adhesion of fibrous tissue and endometrial tissue, which may eventually precipitate into \nchronic inflammation and  infertility3,4.\nEvidence has shown that that the pathogenesis of endometriosis is involved in immune system dysfunction. \nSpecific anti-endometrium antibodies were found in patients with endometriosis, which can explain the pain \nand chronic  inflammation5. Changes in cytokine concentration were also detected in peritoneal and follicular \n fluids6. Anti-nuclear antibodies (ANA) were detected in 18% of patients with endometriosis, while no patients \nexpressed ANA in the control  group7–9.\nSystemic lupus erythematosus (SLE) is an autoimmune disease that affects multiple organs. The prevalence \nof SLE is higher in females than males at a 1:9 ratio. Its incidence is highest in females of early reproductive age, \nwhich is similar to endometriosis.\nRecent studies have revealed a higher prevalence of allergies and autoimmune diseases in patients with \n endometriosis10–12. However, the relationship between endometriosis and SLE has not been studied thoroughly \nand remains unclear. Owing to limited number of cohort studies focusing on this topic, we aimed to investigate \nthe association between endometriosis and the risk of SLE using data from Taiwan’s National Health Insurance \nResearch Database (NHIRD) in this population-based retrospective cohort study.\nOPEN\n1School of Medicine, Chung Shan Medical University, Taichung City 40201, Taiwan. 2Institute of Medicine, Chung \nShan Medical University, Taichung City 40201, Taiwan. 3Division of Allergy, Immunology and Rheumatology, \nDepartment of Medicine, Chung Shan Medical University Hospital, Taichung City 40201, Taiwan. 4School of Health \nPolicy and Management, Chung Shan Medical University, Taichung City 40201, Taiwan. 5Department of Medical \nResearch, Chung Shan Medical University Hospital, Taichung City 40201, Taiwan. 6Department of Family and \nCommunity Medicine, Chung Shan Medical University Hospital, Taichung City 40201, Taiwan. 7Graduate Institute \nof Integrated Medicine, China Medical University, Taichung City 40201, Taiwan.  8 These authors contributed \nequally: Yu-Hsi Fan and Pui-Ying Leong. *email: jccwei@gmail.com\n\n2\nVol:.(1234567890)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\nMethods\nData source. The Taiwanese government established Taiwan’s National Health Insurance (NHI) program in \n1995, which is a nationwide, government-run compulsory insurance program, to provide universal health cover-\nage to citizens and foreigners in Taiwan. In 2014, the program covered over 99.9% of Taiwan’s  population13. Reg-\nistration files and claimed data including registry of patient’s diagnosis, drug prescriptions and received medical \nservice, etc., from the NHI program were collected and sorted into the National Health Insurance Research \nDatabase (NHIRD) through full encryption. The Longitudinal Health Insurance Database (LHID) served as \nthe dataset for this cohort, which yielded a sample of 1 million participants from 23 million beneficiaries of the \nNHIRD.\nAs all personal data in this database had been multiply encrypted, the informed consent was waived by the \napproving committee. This study complied with relevant laws and regulations, and it was approved by the Chung \nShan Medical University Hospital Institutional Review Board (CS15134).\nStudy group and outcome measurement. This research adopted a retrospective cohort study design. \nFirst, female subjects were selected from the one million participants acquired from the LHID. Next, we identi-\nfied the endometriosis group with patients with newly diagnosed endometriosis (The International Classifica-\ntion of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes = 617) from 2000 to 2011 confirmed \nwith gynaecological ultrasonography or laparoscopy. The non-endometriosis group consisted of female partici-\npants who were never diagnosed with endometriosis from 1999 to 2013. In addition, any subject diagnosed with \nSLE before the index date was also excluded.\nThe index date of this cohort was the date of the first endometriosis diagnosis for a patient. The outcome \nmeasurement was the diagnosis of SLE (ICD-9-CM = 710.0) and usage of hydroxychloroquine for one year. The \nend point of this cohort was the occurrence of SLE (December 31st, 2013) or withdrawal from the NHI program, \nwhichever occurred first.\nCovariates and matching. Age, comorbidities, use of corticosteroids, use of NSAIDs and use of hormo-\nnal medications were selected as potential confounders. Previous studies had shown the potential association \nbetween these variables and  SLE14–17. Their associations with endometriosis were demonstrated in the present \nstudy (Table 1). We used the ICD-9-CM codes for identification of the comorbidities of this cohort including \nhypertension (ICD-9-CM codes = 401–405), hyperlipidaemia (ICD-9-CM codes = 272.0–272.4), chronic liver \ndisease (ICD-9-CM code = 571), major depressive disorder (ICD-9-CM codes = 2962, 2963), chronic obstructive \npulmonary disease (ICD-9-CM codes = 490–492, 494, 496), diabetes (ICD-9-CM code = 250), coronary artery \ndisease (ICD-9-CM codes = 410–414), and cerebrovascular disease (ICD-9-CM codes = 430–438). Data on the \nuse of medications mentioned in this study was available from LHID, which included any prescriptions of pre-\nTable 1.  Demographic characteristics of the endometriosis and non-endometriosis groups. Bold font \nrepresents statistical significance (P < 0.05). Propensity score matching (PSM) by age, comorbidities, \ncorticosteroids use, NSAIDs use, and hormonal medications use. SLE Systemic lupus erythematosus, PSM \npropensity score matching, COPD chronic obstructive pulmonary disease, NSAIDs non-steroidal anti-\ninflammatory drugs.\nCharacteristic\nBefore PSM\nP value\nAfter PSM\nP value\nEndometriosis \n(N = 16,578)\nNon-\nendometriosis \n(N = 100,548)\nEndometriosis \n(N = 14,967)\nNon-\nendometriosis \n(N = 29,934)\nn % n % n % n %\nAge 1 0.117\n < 30 3645 21.8 21,870 21.8 3645 21.8 3645 21.8\n30–45 8935 53.3 53,610 53.3 8935 53.3 8778 52.4\n ≥ 45 4178 24.9 25,068 24.9 4178 24.9 4335 25.9\nMean ± SD 38 ± 9.4 38 ± 9.4 1 38 ± 9.4 38 ± 9.4 0.676\nComorbidities\nHypertension 746 4.5 3598 3.6  < 0.001 746 4.5 744 4.4 0.958\nHyperlipidemia 316 1.9 1451 1.4  < 0.001 316 1.9 310 1.8 0.809\nChronic liver disease 308 1.8 1149 1.1  < 0.001 308 1.8 311 1.9 0.903\nMajor depressive disorder 116 0.7 474 0.5  < 0.001 116 0.7 114 0.7 0.895\nCOPD 163 1.0 629 0.6  < 0.001 163 1.0 159 0.9 0.823\nDiabetes 305 1.8 1660 1.7 0.114 305 1.8 295 1.8 0.680\nCoronary artery disease 136 0.8 585 0.6  < 0.001 136 0.8 128 0.8 0.621\nCerebrovascular disease 62 0.4 308 0.3 0.174 62 0.4 58 0.3 0.715\nCorticosteroids 8881 53.0 38,684 38.5  < 0.001 8881 53.0 8881 53.0 1.000\nNSAIDs 11,953 71.3 53,653 53.4  < 0.001 11,953 71.3 11,961 71.4 0.923\nHormonal medications 5292 31.6 11,013 11.0  < 0.001 5292 31.6 5289 31.6 0.972\n\n3\nVol.:(0123456789)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\nscription drugs. Credible diagnoses of endometriosis were defined as an assessment using gynaecological ultra-\nsonography or laparoscopy. Corticosteroids or NSAIDs or hormonal medications use were defined as use for \nmore than 30 days during the observation period. Surgical treatment for endometriosis included laparoscopic \nsurgery, adnexectomy, and hysterectomy. The entire list of surgeries as surgical treatment for endometriosis can \nbe found in Supplementary Table S1.\nFirst, we used a 1:6 age matching for the endometriosis (N  = 16,758) and non-endometriosis groups \n(N = 100,548). Propensity score matching was performed by accounting for covariates including age, comorbidi-\nties, use of corticosteroids, use of NSAIDs, and use of hormonal medications to control confounding between \nthe endometriosis and non-endometriosis groups. Finally, the endometriosis group (N = 16,785) and non-endo-\nmetriosis group (N = 16,785) were identified after matching (Fig. 1).\nStatistical analysis. To compare the baseline characteristics of both groups, a chi-squared test and inde-\npendent t-test were utilized respectively for categorical variables and continuous variables. The cumulative prob-\nability of SLE in the two groups was estimated using the Kaplan–Meier estimator. A log-rank test was then \napplied to test for statistical significance. The Cox proportional-hazards model calculated HR and aHR of SLE for \nthe variables, and 95% CI showed the magnitude and range of them. We further conducted a sensitivity analysis, \nlimiting the endometriosis diagnostic criteria to laparoscopic-diagnosed endometriosis to assess the robustness \nof the results. In subgroup analysis, we separated the participants into subgroups by different variables, includ-\ning age, use of corticosteroids, use of NSAIDs, use of hormonal medications, and surgical treatment to clarify \ntheir effects on the association between endometriosis and SLE. Further, interaction analysis was performed to \nFigure 1.  Description of study design and identification of this cohort study.\n\n4\nVol:.(1234567890)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\ncompare the risks of SLE between different subgroups. A two-tailed P value of < 0.05 was considered statistically \nsignificant. Statistical analyses were performed using SPSS 18.0 (Version 18.0, SPSS Inc., Chicago, USA).\nResults\nBefore matching, for the endometriosis group (N  = 16,578), the number of female patients with hypertension, \nhyperlipidaemia, chronic liver disease, major depressive disorder, chronic obstructive pulmonary disease, coro-\nnary artery disease, corticosteroids use, NSAIDs use, and hormonal medications use was significantly greater \nthan the number of females with these conditions/criteria in the non-endometriosis group (N = 100,548; P < 0.05; \nTable 1). After propensity score matching, there was no significant difference in the baseline factors between \nthe two groups (P > 0.05).\nIn Table 2, the number of SLE cases and person-years, the incidence density rate (per 1000 person-years), \nand crude/adjusted HR were calculated for each category using the Cox proportional-hazards model. In the \nnon-endometriosis group, 18 new cases of SLE occurred during 126,860 person-years, while there were 39 cases \nof new-onset SLE during 113,985 person-years in the endometriosis group. Difference in the incidence density \nrate of SLE was significantly different between the two groups after adjusting for confounders, including age, \ngender, comorbidities, and drug use. The endometriosis group showed a significantly higher incidence density \nrate (0.3 vs. 0.1) and HR/aHR compared with the non-endometriosis group (aHR, 2.37; 95% CI 1.35–4.14). No \nsignificant effects of age, hypertension, chronic liver disease, the uses of corticosteroids, NSAIDs, and hormonal \nmedications were observed in the HR/aHR of SLE with the Cox proportional-hazards model. To evaluate the reli-\nability of the association, we carried out a sensitivity analysis by limiting the endometriosis diagnostic criteria to \nlaparoscopic-diagnosed endometriosis, which is in line with the current gold standard of endometriosis diagnosis \n(Supplementary Table S2). Similar association was demonstrated in the sensitivity analysis. Comparing with the \nnon-endometriosis group, a significant higher aHR for SLE was demonstrated in patients with laparoscopic-\ndiagnosed endometriosis (aHR, 4.74; 95% CI 1.07–20.93).\nTo estimate the difference in the cumulative probability of SLE between the two groups, we used the \nKaplan–Meier estimator and log-rank test to conduct further examination (Fig. 2). In Fig. 2, the endometriosis \ngroup had a significantly higher cumulative probability of new-onset SLE during the study period (P = 0.002).\nTable 3 demonstrates patients with new-onset SLE in both groups were further analysed by using diverse \nvariables (different ages, with or without the use of corticosteroids, NSAIDs or hormonal medications). Interac-\ntion analysis was utilized to compare the risk of SLE across different subgroups. For those who were between \n30 and 45 years of age, patients with endometriosis had an increased risk of SLE (HR, 3.33; 95% CI 1.42–7.81). \nThere were no significant differences in SLE risks between participants less than 30 years of age and greater \nthan or equal to 45 years of age (HR, 1.76; 95% CI 0.57–5.38 and HR, 1.73; 95% CI 0.61–4.86, respectively; P \nvalues for interaction = 0.521). Corticosteroids-free participants (who had never used corticosteroids or used \ncorticosteroids for < 30 days during follow up) who had endometriosis were observed for a higher hazard of SLE \n(HR, 9.29; 95% CI 2.14–40.25). On the other hand, among corticosteroids users (who had used corticosteroids \nfor ≥ 30 days during follow up), patients with endometriosis were not observed for a higher risk of SLE (HR, 1.45; \n95% CI 0.76–2.77; P values for interaction = 0.019). In contrast, a significantly increased risk of SLE for patients \nwith endometriosis was displayed in NSAIDs users (who had used NSAIDs for ≥ 30 days during follow up; HR, \n2.51; 95% CI 1.28–4.94), but not in NSAID-free participants (who had never used NSAIDs or used NSAIDs \nTable 2.  Cox proportional-hazards model analyses HRs and 95% CIs of SLE of each variables. Bold font \nrepresents statistical significance (P < 0.05). Incidence density rate is defined per 1000 person-years. An \nadjusted hazard ratio (HR) is calculated from crude HR after adjusting for age, hypertension, chronic liver \ndisease, corticosteroids use, NSAIDs use, and hormonal medications use. Hyperlipidemia, major depressive \ndisorder, chronic obstructive pulmonary disease, diabetes, coronary artery disease, and cerebrovascular \ndisease were eliminated in the model since only one or no SLE cases were observed. SLE systemic lupus \nerythematosus, HR hazard ratio, NSAIDs non-steroidal anti-inflammatory drugs, CI confidence interval.\nCharacteristics\nNumber of SLE \ncases Person-years\nIncidence density \nrate Crude HR 95% CI Adjusted HR 95% CI\nEndometriosis\nNo 18 126,860 0.1 1 1\nYe s 39 113,985 0.3 2.36 1.35–4.13 2.37 1.35–4.14\nAge (years)\n < 30 13 55,816 0.2 1 1\n30–45 47 196,087 0.2 1.00 0.57–1.73 0.99 0.57–1.73\n ≥ 45 23 79,681 0.3 1.19 0.64–2.23 1.22 0.64–2.31\nComorbidities\nHypertension 2 9307 0.2 0.87 0.21–3.57 0.82 0.19–3.46\nChronic liver disease 2 4559 0.4 1.90 0.46–7.77 1.94 0.47–8.02\nCorticosteroids 38 148,859 0.3 1.34 0.77–2.33 1.72 0.94–3.15\nNSAIDs 40 193,804 0.2 0.63 0.35–1.11 0.54 0.29–1.01\nHormonal medica-\ntions 15 89,857 0.2 0.63 0.35–1.14 0.63 0.34–1.15\n\n5\nVol.:(0123456789)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\nfor < 30 days during follow up; HR, 2.09; 95% CI 0.77–5.66; P values for interaction  = 0.804). As for hormonal \nmedications, a significant higher risk of SLE was observed in the group of hormonal medication-free partici -\npants (who had never used hormonal medications or used hormonal medications for  < 30 days during follow \nup; HR, 2.17; 95% CI 1.14–4.13), but not for hormonal medications users (who had used hormonal medications \nfor ≥  30 days during follow up; HR, 3.03; 95% CI 0.97–9.53; P values for interaction = 0.651).\nIn Table 4, the endometriosis group was divided into two subgroups, the surgical treatment group (those \nwho had received surgical intervention including laparoscopic surgery, adnexectomy, or hysterectomy) and the \nnon-surgical treatment group (those who did not receive any surgical intervention), to investigate whether surgi-\ncal interventions affected the risk of SLE in patients with endometriosis. When being compared with the non-\nendometriosis group, both surgical treatment endometriosis group and non-surgical treatment endometriosis \nFigure 2.  Kaplan–Meier estimator and the log-rank test for cumulative probability of endometriosis and non-\nendometriosis group.\nTable 3.  Subgroup analysis for HRs and 95% CIs of SLE according to different variables. Bold font represents \nstatistical significance (P < 0.05). Hazard ratio (HR) was estimated using the univariate Cox proportional-\nhazards model. SLE Systemic lupus erythematosus, HR hazard ratio, NSAIDs non-steroidal anti-inflammatory \ndrugs.\nCharacteristics\nEndometriosis Non-endometriosis\nHR 95% CIn Number of SLE cases n Number of SLE cases\nAge (years)\n < 30 3645 8 3645 5 1.76 0.57–5.38\n30–45 8935 22 8778 7 3.33 1.42–7.81\n ≥ 45 4178 9 4335 6 1.73 0.61–4.86\nP for interaction = 0.521\nCorticosteroids\nNo 7877 17 7877 2 9.29 2.14–40.25\nYe s 8881 22 8881 16 1.45 0.76–2.77\nP for interaction = 0.019\nNSAIDs\nNo 4805 11 4797 6 2.09 0.77–5.66\nYe s 11,953 28 11,961 12 2.51 1.28–4.94\nP for interaction = 0.804\nHormonal medications\nNo 11,466 28 11,469 14 2.17 1.14–4.13\nYe s 5292 11 5289 4 3.03 0.97–9.53\nP for interaction = 0.651\n\n6\nVol:.(1234567890)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\ngroup had a higher risk in SLE (aHR 1.61; 95% CI 0.59–4.36 and aHR 2.55; 95% CI 1.44–4.51, respectively). \nBy further comparing the surgical treatment endometriosis group and non-surgical treatment endometriosis \ngroup, no statistically significant difference was found between their risks of SLE (aHR 0.61; 95% CI 0.24–1.58).\nDiscussion\nThe associations between endometriosis and SLE remain obscure due to limited number of relevant cohort \nstudies. One of the earliest research studies was conducted by the Endometriosis Association in 1998. A cross-\nsectional research study featuring 5500 female patients from the USA showed a prevalence odds ratio of 20.7 \n(95% CI 14.3–20.9) for  SLE18. Nonetheless, this study was limited by its highly self-driven participants and the \nlack of adjustment for confounders in the statistical analysis. An elevated risk of SLE was confirmed by a 22-year \nfollow-up study featuring nurses with 6434 patients laparoscopically diagnosed with endometriosis (HR 2.03; \n95% CI 1.17–3.51)19. Recently, this correlation was also reported by a case–control study using Swedish registers \n(odds ratio 1.39; 95% CI 1.09–1.78) 20. The advantage of this research was its substantial follow-up time, but the \nlack of matching for specific variables, such as comorbidities, drugs, etc., had become a limitation of this study.\nA Danish cohort study did not support the same association as above. This study analysed 9191 patients whose \nendometriosis were diagnosed by laparoscopy or  laparotomy21. There were no significant associations between \nendometriosis and SLE (standardized incidence ratio = 1.1; 95% CI 0.6–2.1). However, a modest increase in the \nrisk of SLE was mentioned in the study.\nThe present study is one of the largest cohort studies focusing on the association between endometriosis and \nSLE in patients from the LHID—a nationwide database from Taiwan with over one million participants. This \nstudy reports a significant association between endometriosis and SLE (aHR 2.37; 95% CI 1.35–4.14). By limiting \nthe endometriosis diagnostic criteria to laparoscopic-diagnosed endometriosis, this association still remains. In \naddition, patients with endometriosis who were corticosteroid-free, NSAIDs-using, hormonal medications-free, \nor between the age of 30 to 45 have a greater risk in developing SLE. Furthermore, for patients with endometrio-\nsis, regardless of whether they received surgical treatment, there was no significant difference in the risk of SLE.\nAberration of the immune system could explain the correlation between endometriosis and SLE. Several \npieces of evidence imply that an overactive adaptive immune system may play a significant role in the patho -\ngenesis of endometriosis and SLE. ANA was detected in 18% of patients with endometriosis, 100% of patients \nwith SLE, and was not detected in the control group in a prospective randomized  trial7. In addition, intense B \ncell activation and tumour necrosis factor-α (TNF-α) upregulation were observed in both the endometriosis \nand SLE  groups22,23. Characteristics of underactivity in innate immune cells were also observed in the SLE \nand endometriosis groups. A recent study demonstrated that neutrophils had a decreased rate of apoptosis in \npatients with endometriosis compared with disease-free  females24,25. As for patients with SLE, the dysfunctional \nphagocytic ability of neutrophils is well-acknowledged and contributes to vascular damage, lupus nephritis, and \nskin disease in patients with  SLE26,27. Although it is still unclear how the abnormal expression of the immune \nsystem participates in the pathogenesis of endometriosis and SLE, this would be a reasonable explanation for \nthe association between endometriosis and SLE.\nThe dysfunction of cytokine networks and complementary systems could be another mechanism linking \nendometriosis and SLE. Changes in cytokine concentrations in peritoneal and follicular fluids were discovered \nin patients with endometriosis, and a higher level of mannan-binding lectin serine protease 1 (MASP-1) was \nfound both in patients with endometriosis and in patients with  SLE6,28,29. MASP-1 has various functions, includ-\ning activation of the complement and cytokine system, and regulation of endothelial cell  function30. Moreover, a \nrecent animal study evidenced the involvement of MASP-1 in the development of lupus-like glomerulonephritis \nin  mice31. However, our current understanding of MASP-1 still cannot prove whether a higher amount of MASP-1 \nis related to the pathogenesis of endometriosis or SLE.\nHormones, especially oestrogen, had been proven playing an important role in the development of both endo-\nmetriosis and  SLE2,32. The incidence of the two diseases increases after menarche and decreases after menopause. \nOverexpression of oestrogen receptor α (ERα) mRNA was found in endometrial tissue, which is related to the \ndevelopment and growth of endometrial  tissue33. Recent research has also reported the exaggerated response of \npatients with SLE to  oestrogen34. However, it is still largely unknown how oestrogen participates in the emergence \nof endometriosis and SLE. Our study reported that endometriosis significantly increased the risk of SLE among \nhormonal medications-free participants. Even higher risk of SLE was observed in hormonal medications users, \nTable 4.  Cox proportional-hazards model analyses HRs and 95% CIs of SLE for patients with endometriosis \nwith surgical treatment. Bold font represents statistical significance (P < 0.05). Hazard ratio (HR) was estimated \nusing the univariate Cox proportional-hazards model. SLE Systemic lupus erythematosus, HR hazard ratio.\nCharacteristics N Number of SLE cases Crude HR 95% CI Adjusted HR 95% CI\nEndometriosis\nNo 16,758 18 1 1\nNon-surgical treatment 13,678 34 2.58 1.45–4.57 2.55 1.44–4.51\nSurgical treatment 3080 5 1.51 0.56–4.06 1.61 0.59–4.36\nEndometriosis alone\nNon-surgical treatment 13,678 34 1 1\nSurgical treatment 3080 5 0.60 0.23–1.52 0.61 0.24–1.58\n\n7\nVol.:(0123456789)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\nalthough there was no statistical significance in this association. Smaller sample size in the subgroup and the \nexcessively broad classification of hormonal medications may be the reason. Because of the scarcity and incon-\nsistency of the results, studies with larger sample sizes and more detailed classification of hormonal medications \nare still needed for in-depth exploration of this finding.\nSurgical treatment is usually recommended for patients if the endometriosis is deep, causes severe pain or \naffects fertility. However, the correlation between the surgical management and the incidence of SLE is still \nuncertain. A case–control study focusing on European-Americans and African-Americans revealed a signifi-\ncantly increased rate of hysterectomy among patients with  SLE35. However, another case–control study in the \nUnited States demonstrated the protective effect of prior hysterectomy (odds ratio 0.55; 95% CI 0.31–0.99) 36. \nAccording to our results in Table  4, the non-surgical treatment group had the highest risk of SLE among the \nthree subgroups (aHR 2.55; 95% CI 1.44–4.51). However, when focusing on the comparison of surgical treatment \ngroup and non-surgical treatment group, there was no statistical significance between their risks of SLE. Similar \nresults were presented in another Taiwanese cohort study, showing the limited influence over the risk of SLE for \nendometriosis patients under surgical  treatment37. It is worth noticing that owing to the lack of information on \nthe severity of patients’ endometriosis in our dataset, we were unable to adjust for this variable. More large-scale, \ncarefully designed studies in the future are needed to clarify the effect of surgical intervention on the risk of SLE.\nOur study is noteworthy because we found that with the use of corticosteroids, female participants who had \nendometriosis did not have a significant risk of SLE (HR 1.45; 95% CI 0.76–2.77). In contrast, corticosteroids-\nfree participants who had endometriosis were observed to have a significantly higher HR of SLE as 9.29 (95% CI \n2.14–40.25). Similar study by Lin, et al. did not stratified patients with the use of glucocorticoids or  NSAIDs37. \nIn fact, corticosteroids have been commonly used to treat SLE due to their anti-inflammatory and immuno-\nsuppressive effects. Corticosteroids effectively ablate the inflammation by suppressing the activity of cyclooxy-\ngenase enzymes (COX-1 and COX-2) and downstream proinflammatory factors such as prostaglandin and \n thromboxane38. Corticosteroids are also widely recognized for their immunosuppressive actions by repressing \nmultiple immunomodulatory transcription factors. Previous researches revealed a long-term exogenous corticos-\nteroids exposure may induce corticosteroid resistance, which will lead to increased susceptibility to autoimmune \n diseases39. However, the present study did not demonstrate similar results in patients with endometriosis. Our \nspeculation to our results is that when received corticosteroids, the patient’s autoimmune response was sup-\npressed by corticosteroids, attenuating the additional risk of SLE for endometriosis patients. The association was \nnot observed in participants using NSAIDs. This can be explained since corticosteroids not only have a stronger \nanti-inflammatory effect than NSAIDs, but also act as immunosuppressive agents. Further understanding about \nthe pathogenesis of SLE and more careful studies on corticosteroids are necessary to confirm our speculation.\nIt needs to be further explained that drugs are classified into prescription drug, indicator drug and over-the-\ncounter (OTC) drug in Taiwan. The latter two can be purchased without prescriptions from physicians, which \nmeans that our participants may purchase some specific types of drug such as hormone or contraceptives or \nOTC NSAIDs from the pharmacies outside the record of our database. Except for few types of oral gel and topical \nointments, corticosteroids are all prescription drugs. As a result, the lack of complete records on participants’ use \nof NSAIDs, hormones treating for endometriosis, glucocorticoids may also confound the results. However, the \ncoverage of the NIH programme is 99% of the population and patients in Taiwan can get easy access to medical \ncare could reduce the cofounding.\nThe limitations of our research should be highlighted. First, we were unable to acquire laboratory data that \ncould monitor participants’ immune systems or hormone levels to further validate our hypothesis as this is a \nretrospective cohort registration study. Second, some risk factors, such as smoking, alcohol consumption, body \nmass index, and family history were not recorded in our  database40. These variables may confound our discov-\nered associations between endometriosis and SLE to some degree. Besides, both SLE and endometriosis are \ndiseases that are often delayed in diagnosis due to their atypical onset symptoms. Also, they both share similar \nage-specific incidence. As a result, even patients diagnosed with SLE before index date were all excluded, it is \nstill possible that a small number of patients with undiagnosed SLE were enrolled in the cohorts, which could \ninfluence the association. In addition, certain subgroup analyses may be limited by the small number of cases, \nwhich may increase the chance of erroneously obtaining subgroup effects and interactions. Moreover, given the \nracial differences in the incidence of systemic lupus erythematosus, and the lack of diversity in ethnicities in \nTaiwan population, the results of this study may not be generalized for other races/ethnicities. Finally, patients \nin the endometriosis group might have had a greater number of medical visits and therefore a greater chance of \nbeing diagnosed with SLE, which would represent detection bias.\nOur research possesses several major advantages. First, as mentioned above, the use of data from large-scale, \nnationwide, randomly assigned samples from the LHID/NHIRD could reduce selection bias. Second, all claim \ndata in LHID/NHIRD was carefully reviewed by experts and clinicians to ensure validity. Finally, although SLE \nhas a gradual onset and is commonly underdiagnosed for years, a 12-year follow-up in our study would be suf-\nficient to minimize underestimation of the incidence of SLE.\nIn conclusion, we report a significant association between endometriosis and SLE. However, further basic \nmedical researches are needed to clarify the link and mechanisms between endometriosis and SLE.\nData availability\nAll data relevant to the study are available from the National Health Insurance Research Database (NHIRD), \nwhich is provided by the National Health Insurance (NHI) administration, Ministry of Health and Welfare \nof Taiwan and the National Health Research Institutes (NHRI) of Taiwan. It is not publicly available because \nit restricts only the researchers or clinicians who applied and signed an agreement with NHRI are eligible to \napply for the National Health Insurance Research Database (NHIRD). The following is the official website of the \nNHIRD (https ://nhird .nhri.org.tw/).\n\n8\nVol:.(1234567890)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\nReceived: 27 May 2020; Accepted: 14 December 2020\nReferences\n 1. Eskenazi, B. & Warner, M. L. Epidemiology of endometriosis. Obstet. Gynecol. Clin. N. Am. 24, 235–258 (1997).\n 2. Vercellini, P ., Viganò, P ., Somigliana, E. & Fedele, L. Endometriosis: pathogenesis and treatment. Nat. Rev. Endocrinol. 10, 261. \nhttps ://doi.org/10.1038/nrend o.2013.255 (2013).\n 3. Burney, R. O. & Giudice, L. C. Pathogenesis and pathophysiology of endometriosis. Fertil. Steril.  98, 511–519. https ://doi.\norg/10.1016/j.fertn stert .2012.06.029 (2012).\n 4. Sampson, J. A. Peritoneal endometriosis due to the menstrual dissemination of endometrial tissue into the peritoneal cavity. Am. \nJ. Obstet. Gynecol. 14, 422–469. https ://doi.org/10.1016/S0002 -9378(15)30003 -X (1927).\n 5. Mathur, S., Butler, W . J., Chihal, H. J., Isaacson, K. B. & Gleicher, N. Target antigen(s) in endometrial autoimmunity of endome-\ntriosis. Autoimmunity 20, 211–222 (1995).\n 6. Kyama, C. M., Debrock, S., Mwenda, J. M. & D’Hooghe, T. M. Potential involvement of the immune system in the development \nof endometriosis. Reprod. Biol. Endocrinol. RB&E 1, 123. https ://doi.org/10.1186/1477-7827-1-123 (2003).\n 7. Pasoto, S. G. et al. Endometriosis and systemic lupus erythematosus: a comparative evaluation of clinical manifestations and \nserological autoimmune phenomena. Am. J. Reprod. Immunol. 53, 89. https ://doi.org/10.1111/j.1600-0897.2005.00252 .x (2005).\n 8. Vlagea, A. et al. Antinuclear antibodies and cancer: a literature review. Crit. Rev. Oncol. Hematol. 127, 42. https ://doi.org/10.1016/j.\ncritr evonc .2018.05.002 (2018).\n 9. Solomon, D. H., Kavanaugh, A. J. & Schur, P . H. Evidence-based guidelines for the use of immunologic tests: antinuclear antibody \ntesting. Arthritis Rheum. 47, 435. https ://doi.org/10.1002/art.10561  (2002).\n 10. Caserta, D., Mallozzi, M., Pulcinelli, F . M., Mossa, B. & Moscarini, M. Endometriosis allergic or autoimmune disease: pathogenetic \naspects—a case control study. Clin. Exp. Obstet. Gynecol. 43, 355–357 (2016).\n 11. Kvaskoff, M. et al. Endometriosis: a high-risk population for major chronic diseases?. Hum. Reprod. Update 21, 507–509. https ://\ndoi.org/10.1093/humup d/dmv01 3 (2015).\n 12. Soliman, A. M. et al. Comorbidities and symptoms among endometriosis patients: a systematic literature review. Value Health 19, \nA172. https ://doi.org/10.1016/j.jval.2016.03.1445 (2016).\n 13. Administration, N. H. I. National Health Insurance Annual Report 2014–2015 (Ministry of Health and Welfare, Taiwan, 2014).\n 14. Kuo, C. F . et al. Temporal relationships between systemic lupus erythematosus and comorbidities. Rheumatology (Oxford) 58, \n840–848. https ://doi.org/10.1093/rheum atolo gy/key33 5 (2019).\n 15. Shin, S. Drug utilization and therapy provision patterns by prescriber types among patients with systemic lupus erythematosus in \nKorea. Patient Prefer Adherence 11, 1779–1787. https ://doi.org/10.2147/PPA.S1461 19 (2017).\n 16. Chen, H. H. et al. Risk of herpes zoster in patients with systemic lupus erythematosus: a three-year follow-up study using a nation-\nwide population-based cohort. Clinics (Sao Paulo) 66, 1177–1182. https ://doi.org/10.1590/s1807 -59322 01100 07000 09 (2011).\n 17. Askanase, A. D. Estrogen therapy in systemic lupus erythematosus. Treat. Endocrinol.  3, 19–26. https ://doi.org/10.2165/00024  \n677-20040 3010-00003  (2004).\n 18. Sinaii, N., Cleary, S. D., Ballweg, M. L., Nieman, L. K. & Stratton, P . High rates of autoimmune and endocrine disorders, fibromy-\nalgia, chronic fatigue syndrome and atopic diseases among women with endometriosis: a survey analysis. Hum. Reprod. (Oxford, \nEngland) 17, 2715–2724 (2002).\n 19. Harris, H. R. et al. Endometriosis and the risks of systemic lupus erythematosus and rheumatoid arthritis in the Nurses’ Health \nStudy II. Ann. Rheum. Dis. 75, 1279–1284. https ://doi.org/10.1136/annrh eumdi s-2015-20770 4 (2016).\n 20. Harris, H. R., Simard, J. F . & Arkema, E. V . Endometriosis and systemic lupus erythematosus: a population-based case–control \nstudy. Lupus 25, 1045–1049. https ://doi.org/10.1177/09612 03316 63163 5 (2016).\n 21. Nielsen, N. M., Jorgensen, K. T., Pedersen, B. V ., Rostgaard, K. & Frisch, M. The co-occurrence of endometriosis with multi-\nple sclerosis, systemic lupus erythematosus and Sjogren syndrome. Hum. Reprod. (Oxford, England) 26, 1555–1559. https  ://doi.\norg/10.1093/humre p/der10 5 (2011).\n 22. Lisnevskaia, L., Murphy, G. & Isenberg, D. Systemic lupus erythematosus. The Lancet  384, 1880. https ://doi.org/10.1016/S0140 \n-6736(14)60128 -8 (2014).\n 23. Keenan, J. A., Chen, T. T., Chadwell, N. L., Torry, D. S. & Caudle, M. R. IL-1 beta, TNF-alpha, and IL-2 in peritoneal fluid and \nmacrophage-conditioned media of women with endometriosis. Am. J. Reprod. Immunol. 34, 381–385 (1995).\n 24. Ahn, S. H. et al. Pathophysiology and immune dysfunction in endometriosis. Biomed. Res. Int. 2015, 12. https ://doi.\norg/10.1155/2015/79597 6 (2015).\n 25. Kwak, J. Y ., Park, S. W ., Kim, K. H., Na, Y . J. & Lee, K. S. Modulation of neutrophil apoptosis by plasma and peritoneal fluid from \npatients with advanced endometriosis. Hum. Reprod. (Oxford, England) 17, 595–600 (2002).\n 26. Brandt, L. & Hedberg, H. Impaired phagocytosis by peripheral blood granulocytes in systemic lupus erythematosus. Scand. J. \nHaematol. 6, 348–353 (1969).\n 27. Kaplan, M. J. Neutrophils in the pathogenesis and manifestations of SLE. Nat. Rev. Rheumatol. 7, 691–699. https ://doi.org/10.1038/\nnrrhe um.2011.132 (2011).\n 28. Suryawanshi, S. et al. Complement pathway is frequently altered in endometriosis and endometriosis-associated ovarian cancer. \nClin. Cancer Res. 20, 6163–6174. https ://doi.org/10.1158/1078-0432.Ccr-14-1338 (2014).\n 29. Troldborg, A. et al. Levels in plasma of the serine proteases and associated proteins of the lectin pathway are altered in patients \nwith systemic lupus erythematosus. J. Rheumatol. 42, 948–951. https ://doi.org/10.3899/jrheu m.14116 3 (2015).\n 30. Megyeri, M. et al. Complement protease MASP-1 activates human endothelial cells: PAR4 activation is a link between complement \nand endothelial function. J. Immunol. 183, 3409–3416. https ://doi.org/10.4049/jimmu nol.09008 79 (2009).\n 31. Machida, T. et al. Essential roles for mannose-binding lectin-associated serine protease-1/3 in the development of lupus-like \nglomerulonephritis in MRL/lpr mice. Front. Immunol. 9, 1191–1201. https ://doi.org/10.3389/fimmu .2018.01191  (2018).\n 32. Tsokos, G. C. Systemic lupus erythematosus. N. Engl. J. Med. 365, 2110–2121. https ://doi.org/10.1056/NEJMr a1100 359 (2011).\n 33. Matsuzaki, S. et al. Characterization of messenger RNA expression of estrogen receptor-α and -β in patients with ovarian endo -\nmetriosis. Fertil. Steril. 73, 1219–1225. https ://doi.org/10.1016/S0015 -0282(00)00527 -6 (2000).\n 34. Grimaldi, C. M. Sex and systemic lupus erythematosus: the role of the sex hormones estrogen and prolactin on the regulation of \nautoreactive B cells. Curr. Opin. Rheumatol. 18, 456–461. https ://doi.org/10.1097/01.bor.00002 40354 .37927 .dd (2006).\n 35. Namjou, B. et al. The effects of previous hysterectomy on lupus. Lupus 18, 1000–1005. https ://doi.org/10.1177/09612 03309 10431 \n5 (2009).\n 36. Grimes, D. A., LeBolt, S. A., Grimes, K. R. & Wingo, P . A. Systemic lupus erythematosus and reproductive function: a case–control \nstudy. Am. J. Obstet. Gynecol. 153, 179–186. https ://doi.org/10.1016/0002-9378(85)90108 -5 (1985).\n 37. Lin, Y .-H., Y ang, Y .-C., Chen, S.-F ., Hsu, C.-Y . & Shen, Y .-C. Risk of systemic lupus erythematosus in patients with endometriosis: \na nationwide population-based cohort study. Arch. Gynecol. Obstet. https ://doi.org/10.1007/s0040 4-020-05726 -9 (2020).\n 38. Albert, D. A., Hadler, N. M. & Ropes, M. W . Does corticosteroid therapy affect the survival of patients with systemic lupus ery-\nthematosus?. Arthritis Rheum. 22, 945–953 (1979).\n\n9\nVol.:(0123456789)Scientific Reports |          (2021) 11:532  | https://doi.org/10.1038/s41598-020-79954-z\nwww.nature.com/scientificreports/\n 39. Silverman, M. N. & Sternberg, E. M. Glucocorticoid regulation of inflammation and its functional correlates: from HPA axis to \nglucocorticoid receptor dysfunction. Ann N Y Acad Sci 1261, 55–63. https ://doi.org/10.1111/j.1749-6632.2012.06633 .x (2012).\n 40. Hsieh, C. Y . et al. Taiwan’s National Health Insurance Research Database: past and future. Clin. Epidemiol. 11, 349–358. https  ://\ndoi.org/10.2147/clep.S1962 93 (2019).\nAcknowledgements\nThe authors would like to thank the Chung Shan Medical University DryLab Team.\nAuthor contributions\nY .-H.F . and P .-Y .L. wrote the main manuscript text. J.-Y .C. and J.C.-C.W . conceptualized ideas and designed the \nstudy. Y .-H.W . analysed and visualized the data. P .-Y .L., M.-H.K. and J.C.-C.W . edited the manuscript. All authors \nreviewed the manuscript. Y .-H.F . and P .-Y .L. contributed equally.\nFunding\nThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-\nprofit sectors.\nCompeting interests \nThe authors declare no competing interests.\nAdditional information\nSupplementary Information The online version contains supplementary material available at https ://doi.\norg/10.1038/s4159 8-020-79954 -z.\nCorrespondence and requests for materials should be addressed to J.C.-C.W .\nReprints and permissions information is available at www.nature.com/reprints.\nPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and \ninstitutional affiliations.\nOpen Access  This article is licensed under a Creative Commons Attribution 4.0 International \nLicense, which permits use, sharing, adaptation, distribution and reproduction in any medium or \nformat, as long as you give appropriate credit to the original author(s) and the source, provide a link to the \nCreative Commons licence, and indicate if changes were made. The images or other third party material in this \narticle are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the \nmaterial. If material is not included in the article’s Creative Commons licence and your intended use is not \npermitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from \nthe copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.\n© The Author(s) 2021","source_license":"CC0","license_restricted":false}