Introduction

Serological activity in systemic lupus erythematosus (SLE) generally refers to the presence of anti-dsDNA antibodies and/or hypocomplementemia.1 A subset of SLE patients, first described by Gladman et al in 1979, has been recognised for their persistent serological activity despite clinical quiescence.2 3 While some patients remain serologically active clinically quiescent (SACQ) for long durations or become serologically quiescent clinically quiescent (SQCQ), others’ SACQ periods are terminated by a disease flare, for which reliable predictors have yet to be identified.3 4 An earlier study reported that SLE patients with a prolonged SACQ period had comparable disease-related damage to SQCQ patients and suggested the practice of active surveillance without treatment with glucocorticoids (GCs) or immunosuppressives during the SACQ period.5 This and other initial studies excluded patients on GCs or immunosuppressive medications from SACQ. However, the current definition of remission in SLE (DORIS) allow prednisolone≤5 mg/day, immunosuppressives, and biologics,6 and lupus low disease activity state (LLDAS), which allows prednisolone dosage≤7.5 mg/day, has been validated as an endpoint for clinical studies in SLE.7 A more recent retrospective study on SACQ from a Chinese hospital permitted≤7.5 mg/day of prednisone and immunosuppressants and demonstrated that SACQ patients experienced flares more frequently than SQCQ patients (33% vs 16%, p<0.001).8 On the other hand, lupus disease activity may need to be assessed while accounting for GC doses. In order to more accurately gauge the severity of disease activity, investigators of the University of Toronto Lupus Clinic have developed a novel modification of Systemic Lupus Erythematous Disease Activity Index 2000 (SLEDAI-2K), SLEDAI-2K GC (SLEDAI-2KG) by adding weighted scores for GC doses.9 For instance, SLEDAI-2K of 6 in a patient on 10 mg/day of prednisone (=GC weighted score of 3) results in SLEDAI-2KG of 9. In this context, patients using >7.5 mg/day of prednisone may not be regarded as ‘clinically quiescent’, even if their clinical SLEDAI-2K scores are zero. Thus, appropriate GC therapy in SACQ patients with SLE (SACQ-SLE) including those receiving low-dose GC, such as≤7.5 mg/day of prednisolone-equivalent GCs, should be further investigated.

Long-term GC therapy should be minimised so as to avoid various detrimental effects including irreversible organ damage.10 However, a meta-analysis on risk of flare associated with GC withdrawal in SLE (including but not limited to SACQ) demonstrated that SACQ was associated with an increased risk of flare after GC withdrawal (OR 1.78, 95% CI, 1.00 to 3.15).11 An international survey of 130 clinicians reported that more than 40% of the clinicians preferred to continue low-dose GC (prednisolone 5 mg/day) in SACQ patients even when the duration of clinical remission was 5 years.12 However, a number of SACQ patients and outcome events (flares) in the previous studies were not sufficient in size to draw conclusions. In order to better understand how to manage SACQ-SLE, we examined the risk of flare and damage accrual after tapering GCs and also the association between the use of other medications and flare in SACQ-SLE using prospectively collected data from a large, multinational SLE patient cohort. Originally, persistency was one of the components defining SACQ,2 but the required duration varied between studies3 5 8: some studies did not consider duration11 as in remission6 and LLDAS.7 For this study, duration was not considered as the criterion of SACQ, and a term ‘modified SACQ (mSACQ)’ is used throughout this study to mean this ‘SACQ not-considering duration’ to distinguish it from the original definition of SACQ.

MethodsStudy population

Study participants were selected from the Asia Pacific Lupus Collaboration (APLC) patient cohort,13 which included patients recruited from 24 sites across 12 countries, and followed between 2013 and 2020. All patients were consenting adults who met either the 1997 American College of Rheumatology (ACR) modified classification criteria for SLE14 or the Systemic Lupus International Collaborating Clinics (SLICC) 2012 classification criteria.15 The study was restricted to patients who met the mSACQ definition (as defined below) at least once during the study observation period (2013–2020) and with 2 years of subsequent follow-up data since the first mSACQ visit.

Patient and public involvement

Patients or the public were not involved in the design or conduct of this study.

Data collection

Data were collected prospectively during routine patient follow-up using standardised paper or electronic forms for data collection. Baseline demographic data, such as age, gender, self-reported ethnicity and date of SLE onset, were collected at enrolment in the APLC patient cohort. The minimum prescribed visit frequency was 6 months, with the majority of patients having more frequent visits based on clinical need. Disease activity (SLEDAI-2K),16 Safety of Estrogens in Lupus Erythematosus National Assessment (SELENA)-SLEDAI flare index17 and Physician Global Assessment18 and data on all medications and doses were collected at each visit. Doses of GCs were converted into and expressed as prednisolone equivalent. Organ damage was measured at baseline and annually using the SLICC/ACR damage index (SDI).16 A change of one unit in SDI has been demonstrated to be clinically significant19 and was chosen to define damage accrual.

Definitions of mSACQ

mSACQ was defined as the state with serological activity (an SLEDAI-2K Score of 2 or 4 from positive anti-dsDNA antibody and/or hypocomplementemia) but without clinical activity (clinical SLEDAI-2K=0) at any point in time. The original definition of SACQ required persistency, such as 2-year period, and being off-treatment (including GCs and immunosuppressants, but not including antimalarials).5 However, as described in the Introduction section, duration was not considered as the criterion of mSACQ in this study. Patients who were treated with 0 to 7.5 mg/day of prednisolone or equivalent GC on the same visit where patients were on mSACQ state were included in this study (patients not taking GCs were also included). Antimalarials and/or immunosuppressives including biologics were also allowed because they were permitted in the DORIS6 and LLDAS.7 Only the first visit when a patient fulfilled the mSACQ criteria during the observation period was considered as her/his baseline visit for this study.

Outcomes

The primary outcome of our study was disease flare. Flares were divided into severe flares and overall flares, which included mild-to-moderate flares and severe flares, according to the SELENA-SLEDAI flare index definitions.17 In addition, renal flare, which was a component of severe flare and defined as new/worse nephritis, was independently analysed as an exploratory outcome. Accrual of irreversible organ damage (any increase in SDI Score) was the secondary outcome. In addition, diabetes, osteoporosis and avascular necrosis based on individual SDI items were studied because these specific organ damage domains are known to be associated with GC use.20

Data sharing statement

Access to APLC pooled data is subject to the specific guidelines outlined in the APLC Data Access Policy (available on request). The APLC welcomes requests for aggregate (summary) data or to perform analyses of new research questions, and such requests can be submitted to the APLC Steering Committee via the APLC Project Manager.

Statistical analyses

Continuous variables were described as mean (SD). Categorical variables were described as count and proportion (%). As visit-level analyses, Cox proportional hazard models were used to assess the association between decreasing GC dosages (per 1 mg decrease of prednisolone) and time-to-first flare at each subsequent visit, or time-to-first subsequent damage accrual (≥1-point increase in SDI) (recurrent flare/damage accrual events were not counted). In addition, SLE disease-specific and demographic variables, namely initial prednisolone dosage, antimalarial use, immunosuppressive use, disease duration, SLEDAI-2K, age at visit, gender and ethnicity, were included as covariates in the Cox proportional hazard models referencing those of previous studies.7 20 The proportional hazards assumption was assessed by the log of time interaction terms. In the analyses using Cox proportional hazard models, we made separate analyses for each initial dosages of prednisolone-equivalent GC as follows: 0≤ and ≤5 (relevant to remission), 0≤ and ≤7.5 (relevant to LLDAS) and 5<and ≤7.5 (relevant to LLDAS but not in remission). Kaplan-Meier plots of disease flare-free survival with and without antimalarial use were also analysed. The observation period for the survival analyses began from the first visit when each patient fulfilled the mSACQ definition during the observation period and continued for 2 years or censored when each event (flare or increase in SDI Score) occurred for the first time within 2 years. The observation was not censored when patients transitioned to SQCQ. In the exploratory analyses, diabetes, osteoporosis and avascular necrosis were independently analysed as an SDI item, and only decreases in the prednisolone dosage and initial prednisolone dosages were analysed as risk factors due to the small numbers of each event. Renal flare was also independently analysed as an exploratory outcome. In addition, in order to address the potential influence of different therapeutic strategies by clinician from different centres, we carried an additional analysis using a Cox frailty survival model with centre of enrolment as random effect. Missing data were handled using available case analysis, that is, each analysis was based on subjects with available data on the included variables. Two-sided p values of <0.05 were considered statistically significant. No correction for multiple comparisons was made, and p values were interpreted as exploratory results, except for the primary evaluation (GC dose reduction in the flare analysis). Statistical analyses were performed using Microsoft Excel software (V.2021; Microsoft, Redmond, Washington, USA), JMP Pro statistical software (V.17.0.0; SAS Institute, Cary, North Carolina, USA) and R (V.4.3.1).

ResultsClinical characteristics of SLE patients

Out of 4106 patients in the APLC patient cohort, 2268 (55.2%) met the study selection criteria: that is, they were in mSACQ state and using prednisolone (or equivalent GC) ≤7.5 mg/day at the same visit at least once during the period. Among these, 1850 patients with 2 years of subsequent follow-up data since the first mSACQ visit were studied (a total of 8905 visits were constructed). In the study cohort, GC dosages were reduced in 411 patients, whereas they were increased in 164 patients during the observation period. The distribution of recorded reduction in prednisolone dosage was as follows: 0.5 mg/day, 5.2%; 1.0 mg/day, 32.3%; 1.5 mg/day, 32.3%; 2.0 mg/day, 2.4%; 2.5 mg/day, 33.7%; 3.0 mg/day, 1.3%; 3.5 mg/day, 2.3%; 4.0 mg/day, 1.4%; 4.5 mg/day, 0.3%; 5.0 mg/day, 13.5%. As detailed in table 1, the mean (SD) age at enrolment in this study (ie, when each patient fulfilled the mSACQ definition for the first time during the observation period) was 40.0 (13.5), 1678 (91.6%) patients were women and 1645 (89.1%) patients were of Asian ethnicity. By the definition of mSACQ, the clinical SLEDAI-2K scores were 0 and the serological items scores were 2 or 4 at enrolment in all the patients. At enrolment, 629 (37.5%) patients had organ damage (ie, SDI≥1) and 1410 (76.2%) patients were treated with GCs with mean (SD) daily prednisolone dosage of 3.7 (2.6) mg. At enrolment, 1361 (73.6%) patients used antimalarials (hydroxychloroquine or chloroquine) and 831 (55.1%) used immunosuppressive drugs, namely methotrexate, azathioprine, mycophenolate mofetil, mycophenolic acid, leflunomide, ciclosporin, cyclophosphamide, tacrolimus, rituximab or belimumab.

Table 1

Patient demographics and disease characteristics of mSACQ patients with SLE at enrolment*

Flare

Among the 1850 eligible patients with mSACQ, 742 and 271 patients experienced overall flare and severe flare, respectively. In these patients, 551/742 (74.3%) and 176/271 (64.9%) of overall and severe flare occurred without decreasing their GC dosages, respectively. Cox proportional hazard models demonstrated that tapering GC was not associated with increased risk of subsequent overall or severe flare in any groups according to initial dosages of GC: each unit (1 mg/day) decrease in prednisolone-equivalent GC dosages resulted in adjusted HRs 1.02 (95% CI, 0.99 to 1.05; p=0.27) and 0.98 (95% CI, 0.96 to 1.004; p=0.11) for overall and severe flare, respectively, in the group with initial prednisolone-equivalent GC dosages of 0–7.5 mg/day (≥0 and ≤7.5) (table 2). By contrast, antimalarial use was associated with decreased risk of overall and severe flare in the groups with initial prednisolone-equivalent GC dosages of 0–7.5 mg/day and 0–5 mg/day (≥0 and ≤5) (figure 1A,B, table 3 and online supplemental table 1). In addition, immunosuppressive use was associated with decreased risk of severe flare but not overall flare in these groups (table 3 and online supplemental table 1). Higher initial GC dosage and SLEDAI-2K were also associated with increased risk of overall and severe flare (table 3 and online supplemental table 1). The exploratory analysis showed that tapering GC was not associated with renal flare in the group with initial prednisolone-equivalent GC dosages of 0–7.5 mg/day (HR, 1.04; 95% CI, 0.99 to 1.10; p=0.098) (online supplemental table 3). When a Cox frailty survival model with centre of enrolment as random effect was performed, it was confirmed that tapering GC was not associated with increased risk of subsequent overall or severe flare. However, antimalarial use and immunosuppressive use were not associated with decreased risk of flare although HRs were similar in both models: HRs for increased flare risk with antimalarial use were 0.777 and 0.882 in the Cox standard and frailty models, respectively (table 4 and online supplemental table 4).

Figure 1

Kaplan-Meier plots of disease flare-free or damage accrual-free survival in modified serologically active clinically quiescent (mSACQ) 0–7.5 mg with and without antimalarial use. Systemic lupus erythematosus (SLE) patients with mSACQ and initial prednisolone-equivalent glucocorticoid dosages of 0–7.5 mg/day (≥0 and ≤7.5) at enrolment were included. Flares were assessed and divided into (A) overall flares including mild-to-moderate flares and severe flares and (B) severe flares, according to the Safety of Estrogens in Lupus Erythematosus National Assessment-SLE Disease Activity Index flare index definitions. (C) Accrual of irreversible organ damage was defined as any increase in Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index Score. Cox proportional hazard models were used to assess the association between antimalarial use (hydroxychloroquine and chloroquine) and disease flares at each subsequent visit, adjusting for other covariates listed in table 3.

Table 2

The risk of subsequent flare or damage accrual per 1 mg decrease of prednisolone in mSACQ patients with SLE assessed using Cox proportional hazard models

Table 3

Association between various factors and subsequent flare or damage accrual in mSACQ patients with SLE treated with ≤7.5 mg/day of prednisolone-equivalent GC assessed using Cox proportional hazard models

Table 4

Association between various factors and subsequent flare or damage accrual in mSACQ patients with SLE treated with ≤7.5 mg/day of prednisolone-equivalent GC assessed using Cox frailty proportional hazard models

Damage accrual

Among the 1677 mSACQ patients whose SDI data were available, 180 patients experienced new damage accrual. The frequencies of SDI items were shown in online supplemental table 5. As shown in table 2, tapering prednisolone-equivalent GC dosages by 1 mg/day was associated with decreased risk of damage accrual in the groups with initial prednisolone-equivalent GC dosages of 0–7.5 mg/day (adjusted HR (95% CI), 0.97 (0.96 to 0.99)) and <5–7.5 mg/day (>5 and ≤7.5) (adjusted HR (95% CI), 0.96 (0.94 to 0.99)) but not 0–5 mg/day (adjusted HR (95% CI), 0.98 (0.95 to 1.01)). The exploratory analyses showed that tapering GC was associated with decreased risk of new occurrence of diabetes (online supplemental table 2). In addition, longer disease duration was associated with increased risk of damage accrual (table 3 and online supplemental table 1). Antimalarial use was not associated with damage accrual (figure 1C, table 3 and online supplemental table 1). When a Cox frailty survival model with centre of enrolment as random effect was performed, protection against damage accrual provided by tapering GCs was confirmed among patients treated with >5 mg/day of prednisolone (table 4 and online supplemental table 4).

Discussion

This study of a large multinational cohort demonstrated that tapering GC was not associated with increased risk of subsequent flare in mSACQ-SLE when HRs were assessed per 1 mg decrease in prednisolone-equivalent GC dosages whereas antimalarial use and immunosuppressive use were associated with decreased flare risk in mSACQ patients. Importantly, tapering GCs protected mSACQ patients treated with >5 mg/day of prednisolone-equivalent GC against damage accrual.

The risk of subsequent flare in mSACQ-SLE was not increased by tapering GC in this study. This may not seem to be in line with the result from the meta-analysis on risk of flare associated with GC withdrawal in SLE (including but not limited to SACQ patients receiving low-dose GCs) reporting that GC discontinuation showed an increased risk of flare compared with GC continuation.20 This may be attributed to the differences in statistical methods: we assessed the risk of flare per 1 mg decrease of prednisolone whereas other studies mostly compared GC discontinuation/withdrawal to GC continuation dichotomously. In addition, this meta-analysis included 203 patients (not limited to SACQ) with flare,20 whereas 742 mSACQ patients experienced flare in the present study. A Chinese single-centre retrospective study on SACQ-SLE reported that GC dosage at the time of flare was rather an independent risk factor for flare in SACQ, whereas GC dosages at the start of SACQ were not statistically different between flare and non-flare patients (GC tapering/withdrawal was not assessed).8 We have previously reported that tapering of GC was associated with flares compared with continuation of therapy in stable SLE.21 However, SLE patients with LLDAS or remission were analysed in that study, whereas this study focused on mSACQ-SLE. In addition, the previous study compared flares after drug tapering versus continuing the same therapy but did not asses the flare risk per one-unit decrease in GC dosage.

Antimalarial use and immunosuppressive use were associated with decreased risk of flare in mSACQ patients, although immunosuppressive use was associated only with decreased risk of severe flare but not overall flare. Similar to our findings, the Chinese single-centre retrospective study demonstrated that antimalarials and immunosuppressants were protective factors for flare in SACQ.8 The meta-analysis on risk of flare associated with GC withdrawal in SLE demonstrated that concomitant use of hydroxychloroquine trended towards decreased risk of flare in patients with SLE after GC withdrawal.11 It was also reported that 26 out of 105 SLE patients who discontinued immunosuppressants due to remission experienced a flare after a median 57 months.22 Maintenance therapy with antimalarials independently protected the patients against flares. These results support a general strategy to withdraw GC first before other immunosuppressive drugs, while continuing long-term hydroxychloroquine.10 However, our data on antimalarials should be interpreted with caution because data on dosages per body weight or blood drug concentrations were not available for analysis.

In our study, tapering prednisolone-equivalent GC dosages by 1 mg/day was associated with 4% decreased risk of damage accrual in the mSACQ patients whose initial prednisolone dosages were >5 mg/day (adjusted HR 0.96, 95% CI 0.93 to 0.99). This is in contrast to the conclusion drawn from a meta-analysis that took into account three studies, but these studies were small in sample size compared with our current report, and included only a total of 47 patients who had experienced damage accrual during the observation periods.20 Based on the results of the present study, GC tapering might be especially recommended in mSACQ patients in whom there may be a concern about new occurrence of diabetes. A study with a longer observation period would be required to clearly demonstrate the benefit for further decreasing the GC dosages from 5 mg/day of prednisolone in mSACQ or other SLE patients.

In addition to the factors described above, a few other potential risk factors of flare in mSACQ were identified. A higher initial GC dosage was associated with increased risk of overall and severe flare. This is probably because patients who were able to be treated with lower initial doses of GCs had milder diseases and lower possibilities of flare. Alternatively, patients who were likely to flare might have already flared before attaining mSACQ while tapering GCs from higher dosages of prednisolone. However, these speculations need to be verified using inception cohorts. Furthermore, a one-unit increase in SLEDAI-2K was also associated with increased risk of overall and severe flare at a subsequent visit. Longer disease duration was associated with higher risk of damage accrual in mSACQ patients. Similarly, a systematic literature review reported that longer disease duration was included in the demographic factors influencing the damage accrual in overall SLE patients in addition to male gender, older age, Afro-Caribbean and Indo-Asian ethnicity.23

The strengths of this study are the large number of mSACQ-SLE patients and that HRs were calculated per 1 mg decrease in prednisolone dosages using Cox proportional hazard models. There is no comparison group for GC taper in this study, whereas a type of statistical analysis that compares decreasing versus non-decreasing may demonstrate increased risk of flare after tapering GCs in mSACQ patients. However, continuous covariates and binary covariates have different meanings, and we considered that evaluating the risk per 1 mg decrease in prednisolone-equivalent GC may be more practical and relevant to the real-world management of mSACQ-SLE.

There are several limitations of this study. First, the research question was designed, and the data were analysed retrospectively, although the data were collected prospectively using standardised data collection forms. Second, although the patterns of medication use were broadly similar to those reported in other cohorts, many participating centres were in health systems with limited access to biologics, the use of which was notably low. Third, the follow-up duration was relatively short, and possibly due to this, significant benefits were not demonstrated in terms of damage accrual among mSACQ patients taking less than 5 mg/day. However, we thought that 2-year observation period was enough to study the flare risk (the primary outcome) and prioritised keeping the number of patients large enough over extending the observation period. Fourth, because data regarding drug adherence or blood drug concentrations were not collected, the possibility that adherence in clinically quiescent patients might have affected the results could not to be excluded. Fifth, because this study was undertaken using a cohort of majority Asian ancestry in the Asia-Pacific region, generalisability needs to be confirmed in other cohorts. However, our Cox proportional hazard models demonstrated that ethnicity (non-Asian vs Asian) was not associated with increased/decreased risk for any outcomes. Sixth, although HRs were calculated per 1 mg decrease in prednisolone dosages using Cox proportional hazard models. More than 1 mg/day reduction steps were made in many patients. Seventh, different effects between reduction from different prednisolone dosages, such as 7.5, 5 and 2.5 mg/day were not directly analysed even though subgroup analyses were conducted in the groups with initial prednisolone-equivalent GC dosages of 0–5 mg/day and <5–7.5 mg/day. Finally, the mSACQ definition used in this study, not-considering duration prior to the first visit of SACQ, was different from the original SACQ, which has usually been defined as ‘at least a 2-year period without clinical activity but with persistent serological activity’.3 5 We acknowledge the stability of the state is an important factor and should be investigated in future studies.

In conclusion, this study of a large multinational cohort indicates that tapering GCs was not associated with an increased risk of subsequent flare in SLE patients who were mSACQ with low daily exposure to GCs (≤7.5 mg/day of prednisolone equivalent). Antimalarial use and immunosuppressive use were associated with decreased risk of flare in mSACQ patients. Tapering GCs provided protection against damage accrual in mSACQ patients treated with >5 mg/day of prednisolone even within the 2-year observation periods. These findings suggest cautious GC tapering is feasible even in mSACQ and can reduce GC burden. However, further studies are needed to explore the best strategies for GC tapering in mSACQ and determine the protective effect of biologics against flare in mSACQ.