Fatigue outcomes with anifrolumab in low-disease-activity SLE: DNA damage in plasmacytoid dendritic cells as a candidate biomarker for responders

Abstract

Objective

Investigate the effect of anifrolumab on fatigue in patients with SLE and low disease activity and identify cellular mechanisms associated with the fatigue response.

Methods

12 SLE patients with significant fatigue (Fatigue Assessment Scale (FAS)¹ ≥22) initiating anifrolumab were followed at baseline, 1 month and 3 months. The primary endpoint was fatigue improvement (FAS reduction ≥4 points). Secondary endpoints included FAS, Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), Generalised Anxiety Disorder-7, Quick Inventory of Depressive Symptomatology Self-Report-16, serum interferon (IFN)-α, type I IFN activity and γH2AX expression in myeloid subsets by flow cytometry.

Results

Median baseline SLEDAI was 4 (IQR 4–7) and median FAS was 28 (IQR 24–36). Fatigue improvement, defined by FAS reduction ≥4 points, was observed in 58% of patients (7/12; 95% CI 29% to 88%) at 3 months. Serum IFN-α was low across the cohort and did not predict response. Baseline γH2AX in plasmacytoid dendritic cells (pDCs) was significantly higher in responders than in non-responders and healthy controls and showed a declining trend after 1 month of treatment (Wilcoxon signed-rank test, n=6, p=0.156).

Conclusions

These preliminary findings suggest that anifrolumab may improve fatigue in patients with SLE including those with low disease activity and low serum IFN-α. Elevated pDC γH2AX at baseline, reflecting subclinical cellular stress, may identify patients who could benefit from IFN blockade for fatigue. These hypothesis-generating findings warrant validation in controlled studies.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Fatigue is prevalent in SLE but it correlates poorly with disease activity and lacks established treatment. Whether interferon (IFN) pathway blockade can alleviate fatigue independently of systemic disease activity has remained unknown.

WHAT THIS STUDY ADDS

• Anifrolumab improved fatigue in 58% of SLE patients with low disease activity and low serum IFN-α. γH2AX in plasmacytoid dendritic cells was elevated at baseline in fatigue responders and decreased after treatment, identifying cellular stress possibly driven by chronic IFN as a candidate predictive marker.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Plasmacytoid dendritic cell γH2AX may predict anifrolumab response in SLE-related fatigue. Fatigue should be treated as an explicit target in SLE management, and IFN blockade warrants consideration in patients with refractory fatigue irrespective of disease activity.

Introduction

SLE is a systemic autoimmune disease that causes inflammation in multiple organs. In addition to its diverse clinical manifestations—arthritis, cutaneous rashes and nephritis—chronic fatigue is among the most prevalent and debilitating symptoms, affecting more than 80% of patients in European cohorts and substantially impairing quality of life.^{1 2} Fatigue is excluded from standard disease activity indices such as SLEDAI and shows no correlation with these scores,^{1 3 4} meaning many patients remain persistently fatigued even in clinical remission.^{5 6} This disconnect between fatigue and measurable disease activity contributes to under-recognition and inadequate management by clinicians.⁷ In addition to fatigue, anxiety and depression also significantly impair the quality of life in patients with SLE.⁸

Type I interferons (IFNs), particularly IFN-α, are central to the pathogenesis of SLE.⁹ Dysregulated IFN responses occur in peripheral blood, inflammatory sites and even non-lesional skin,^10–12 and cells such as plasmacytoid dendritic cells (pDC) and monocytes demonstrate enhanced IFN-α production.^13–15 Patients with elevated IFN signatures frequently present with constitutional symptoms including fatigue,^{16 17} and the expression of the IFN-stimulated gene (ISG) often persists despite clinical remission.¹⁸ Recombinant IFN-α therapy in melanoma caused significant fatigue in 70% of patients at 52 weeks,¹⁹ underscoring the direct fatigue-inducing potential of IFN signalling and suggesting persistent IFN responses may explain the discordance between fatigue and conventional disease activity measures.

Anifrolumab, a human monoclonal antibody against the type I IFN receptor, has demonstrated efficacy in SLE disease activity and IFN-mediated symptoms.^20–23 The trials TULIP-1 and TULIP-2 showed that anifrolumab alleviates fatigue,²⁴ but these trials enrolled patients with relatively high disease activity (~70% had SLEDAI-2K ≥10) and approximately half were receiving prednisolone ≥10 mg. Whether fatigue improvement resulted from enhanced disease control, reduced glucocorticoid burden or a direct effect on IFN-driven pathology therefore remains unclear.

Several lines of evidence implicate DNA damage in SLE. Ultraviolet (UV)-induced oxidative DNA damage drives skin inflammation through the accumulation of modified DNA species such as 8-oxoguanine.^{25 26} Peripheral blood mononuclear cells (PBMCs) from SLE patients exhibit elevated γH2AX, a phosphorylated histone variant that marks DNA double-strand breaks and a key readout of the DNA damage response (DDR).^27–32 In vitro, IFN-α directly induces DDR in healthy B cells,³² suggesting that chronic exposure to IFN drives persistent cellular stress. It remains unclear how these responses change with treatment or relate to clinical manifestations. In this pilot study, we aimed to evaluate the efficacy of anifrolumab on fatigue in SLE patients with low disease activity maintained on a stable glucocorticoid dose and to identify candidate markers predictive of fatigue response.

Methods

Study design and patients

This was a prospective observational study of patients aged ≥18 years of age diagnosed with SLE who initiated anifrolumab between May and December 2024 at the Juntendo University Hospital. There were no restrictions on concomitant medications. Patients who discontinued anifrolumab within 3 months were excluded. All patients met the 1997 American College of Rheumatology classification criteria for SLE. Anifrolumab was administered as 300 mg intravenous infusions every 4 weeks. The activity of the disease was evaluated by SLEDAI.⁴ Demographic characteristics, clinical history, SLEDAI scores and treatment details were retrieved from electronic medical records and anonymised to ensure confidentiality.

Assessment of fatigue, anxiety and depression

Fatigue was evaluated using the Fatigue Assessment Scale (FAS), a validated tool that assesses physical and mental fatigue in patients with autoimmune diseases.^{33 34} The Japanese-language FAS has established reliability and clinical utility.^{35 36} Patients completed the questionnaire at baseline, 1 month and 3 months after the initiation of treatment. FAS scores range from 0 to 40; a score ≥22 indicates significant fatigue.^{33 34} Based on the minimal clinically important difference, a reduction of ≥4 points from baseline was defined as a clinically significant improvement.^{34 37} The patients were classified as the improvement group (decrease ≥4 points) or the non-improvement group.

Anxiety symptoms were assessed using the Generalised Anxiety Disorder 7-item (GAD-7) scale,³⁸ with a threshold score ≥10 indicating probable GAD; a reduction of ≥4 points denoted significant improvement.³⁹ Depressive symptoms were evaluated using the Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR16; total score 0–27). A score ≥6 indicated suspected depression; scores ≤5 were classified as normal.⁴⁰

Blood sample collection and flow cytometry analysis

PBMCs were collected at baseline, 1 month and 3 months after initiating anifrolumab, isolated by density-gradient centrifugation using BD Vacutainer CPT tubes with sodium heparin (BD Biosciences). The samples were stored at −80 °C for 24 hours and then transferred to liquid nitrogen. After thawing, cells were stained with Zombie NIR viability dye (BioLegend #423106) and incubated with FcR blocking reagent (BioLegend). Surface staining used: CD11c-BV785, CD19-BV785, CD14-PerCP/Cy5.5, CD27-PerCP/Cy5.5, CD3/CD11b/CD19/CD56-APC/Cy7, CD15-Alexa700, CD20-Alexa700, IgD-BV421, HLA-DR-BV510 and CD123-BV605 (BioLegend).

For intracellular staining, cells were fixed and permeabilised using the Foxp3/transcription factor staining buffer set (eBioscience #00-5523-00). Anti-H2AX.Phospho(Ser139)-FITC (BioLegend #613404) was used to measure γH2AX. The mean fluorescence intensity (MFI) of γH2AX was quantified in subsets of myeloid cells by flow cytometry.⁴¹ Data were acquired on a BD FACS LSRFortessa and analysed with FlowJo 10.10.0 (Ashland, Oregon, USA).

B cells were defined as CD3⁻CD11b⁻CD19⁻CD200⁺ cells. Memory B cells were identified as CD3⁻CD11b⁻CD19⁺CD20⁺CD27⁺ cells. Double-negative B cells were identified as CD3⁻CD11b⁻CD19⁻CD20⁻CD2727⁻IgD⁻ cells. Naïve B cells were identified as CD3⁻CD11b⁻CD19⁺CD20⁺CD27⁻IgD⁺ cells. The gating strategy is shown in online supplemental figure 1A. Monocytes were defined as CD15⁻CD3⁻CD19⁻CD56⁻CD14⁺ cells. Conventional dendritic cells (cDCs) were identified as CD15⁻CD3⁻CD19⁻CD56⁻CD14⁻HLA-DR⁺CD11c⁺CD123⁻ cells. pDCs were identified as CD15⁻CD3⁻CD19⁻CD56⁻CD14⁻HLA-DR⁺CD11c⁻CD123⁺ cells. The gating strategy is shown in online supplemental figure 1B.

Measurement of serum IFN concentration and activity

The Pre-anifrolumab treatment serum IFN-α concentration was measured by ELISA using the VeriKine-HS Human IFN-α All Subtype ELISA Kit (PBL Assay Science, #41115) according to manufacturer’s instructions.

The reporter cell assay followed established methods.⁴² IFN-α/β reporter HEK293 cells and THP1-dual cells were purchased from InvivoGen (San Diego, California, USA). To quantify IFN-I bioavailability, HEK293 reporter cells were incubated with 10% plasma for 24 hours; secreted embryonic alkaline phosphatase (SEAP) in supernatant was measured using QUANTI-Blue (InvivoGen) at 620 nm.

For IFN-inducing activity, THP1-Dual cells were differentiated by culturing in RPMI-1640 with 10% fetal calf serum and 10 ng/mL phorbol 12-myristate 13-acetate for 24 hours, then rested in RPMI-1640/10% fetal calf serum. Cells were incubated with 5% plasma for 18 hours and Lucia luciferase in supernatant quantified by QUANTI-Luc4 (InvivoGen) chemiluminescence as a readout of interferon-stimulated response element (ISRE)-dependent transcription.

For nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)-activating activity, the same differentiated THP1-dual cells were incubated with 5% plasma for 18 hours; SEAP in the supernatant was quantified by QUANTI-Blue (InvivoGen) at 620 nm as a readout of NF-κB-dependent transcription.

Secondary analysis of TULIP-1/2 transcriptomic data

With permission of the authors of the original study,⁴³ published GSEA results from TULIP-1/2 trials that compared the anifrolumab and placebo groups at week 52 were used. DDR-related gene sets were extracted from the supplemental tables and compiled.

Data visualisation

Data were visualised using Python (V.3.11) with the matplotlib (V.3.1.1). For multi-cell reporter assay comparison (Figure 2), the figures were generated using Python (V.3.11) with the matplotlib (V.3.1.1) and scipy libraries: normalised values of z-score were displayed as a heatmap (RdBu_r colormap; range −2.5 to 2.5 SD) with patients ranked by THP-1 ISRE-luc z-score within each group, correlation analysis as a scatter plot with Spearman’s ρ annotated, and individual z-scores and OR composite score as strip plots. Data organisation and final visualisation of TULIP-1/2 Transcriptomic Data (online supplemental figure 3) were performed using Python (V.3.11) with the pandas (V.1.3.5), numpy (V.1.21.6), matplotlib (V.3.1.1) and seaborn (V.0.12.2) packages.

Study endpoints

The primary endpoint was the fatigue improvement rate. Secondary endpoints included longitudinal changes in FAS, SLEDAI, serum IFN-α concentration and activity and γH2AX expression in myeloid subsets in responders, non-responders and healthy controls.

Statistical analysis

As none of the values followed a normal distribution, two-group comparisons were performed using the Mann-Whitney U test. Spearman’s rank correlation coefficient (two-sided) was used to assess the association between two variables. All γH2AX flow cytometry analyses were pre-specified as exploratory and hypothesis-generating. Differences among three groups were assessed using the Kruskal-Wallis test; when a significant overall difference was identified, post-hoc pairwise comparisons were performed. Bonferroni correction was applied within pairwise comparisons per cell type; no correction was applied across cell types given the pilot nature of this study. Comparisons before and after anifrolumab treatment within the Responder group were performed using the Wilcoxon signed-rank test (paired, two-sided). A p value of <0.05 was considered statistically significant. Calculations were performed by python (V.3.11) and scipy (V.1.13.1).

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Results

Characteristics of the patient and baseline clinical data

A total of 20 patients with SLE initiated anifrolumab between May and December 2024; one was excluded after discontinuation due to a flare of arthritis, and seven were excluded for absence of significant fatigue, yielding a final cohort of 12 patients. No patients discontinued treatment within 3 months due to adverse events. The baseline characteristics are summarised in table 1. The median SLEDAI score was 4 (IQR: 4–7), indicating stable disease activity and the median FAS score was 28 (IQR: 24–36). Since anxiety and depression can coexist with, or drive, fatigue, we assessed these symptoms as potential confounders. At baseline, two patients (16.7%) had GAD-7 scores ≥10 (probable GAD); the median GAD-7 was 5 (IQR: 3–7). Nine patients (75%) had QIDS-SR16 scores ≥6 (suspected depression), including two (16.7%) with scores ≥16 (severe). One patient had a pre-existing diagnosis of anxiety disorder.

Table 1. Baseline characteristics of the patients.

	Entire cohort (n=12)	Fatigue responders (n=7)	Non-responders (n=5)
Age	53 (37–62)	55 (40–68)	39 (34–53)
Disease duration (year)	16 (10–25)	20 (2–23)	15 (12–23)
Female n(%)	11 (91.7)	7 (100)	4 (80)
SLEDAI	4 (4–7)	4 (4–7)	4 (4–6)
Fatigue Assessment Scale	28 (24–36)	28 (26–34)	25 (23–39)
Quick Inventory of Depressive Symptomatology	8 (6–16)	8 (8–10)	7 (2–14)
Generalised Anxiety Disorder-7	5 (3–7)	6 (3–10)	4 (1–6)
Positivity of		n (%)	n (%)
Anti-ds DNA antibody	4 (33.3)	2 (28.6)	2 (40)
Anti-U1-RNP antibody	4 (33.3)	1 (14.3)	3 (60)
Anti-Sm antibody	1 (8.3)	0	1 (20)
Anti-SS-A antibody	6 (50)	2 (28.6)	4 (80)
History of
Belimumab use n(%)	6 (50)	2 (28.6)	4 (80)
Lupus nephritis n(%)	1 (8.3)	0	1 (20)
NPSLE n(%)	0	0	0
Glucocorticoid dosage, mg (Prednisolone equivalent)	6 (5–9)	6 (3.5–10)	5 (4–9)
hydroxychloroquine use	3 (25)	2 (28.6)	1 (20)
Immunosuppressant use	5 (41.7)	2 (28.6)	3 (60)
Symptom:
Arthritis	9 (75)	5 (71.4)	4 (80)
Rash	4 (33.3)	3 (42.9)	1 (20)
Alopecia	2 (16.7)	2 (28.6)	0
Pleurisy	1 (8.3)	1 (14.3)	0
Hypocomplementaemia	3 (25)	1 (14.3)	2 (40)

Numbers are shown by median (IQR）unless otherwise specified.

NPSLE, neuropsychiatric systemic lupus erythematosus; SLEDAI, Systemic Lupus Erythematosus Disease Activity Index.

Early clinical response of fatigue to anifrolumab

Following the introduction of anifrolumab, fatigue improvement, defined by FAS reduction ≥4 points, was observed in 58% of patients (7/12; 95% CI 29% to 88%) at 3 months (figure 1A). Notably, most responders demonstrated a rapid reduction in fatigue as early as 1 month (figure 1B,C), indicating that anifrolumab may induce an early clinical response. Similarly, SLEDAI scores showed an improving trend starting at 1 month (figure 1D–F). Notably, one patient with no SLEDAI improvement showed a response in fatigue (patient 2, figure 1G). In two patients (patients 3 and 5, figure 1G), reduction in FAS preceded improvement in SLEDAI. Prednisolone dose remained unchanged or decreased by no more than 1 mg/day in most patients, with the exception of one patient who was receiving a higher dose of 20 mg/day (figure 1G, online supplemental figure 2A,B). Baseline GAD-7 and QIDS-SR16 scores did not differ significantly between fatigue responders and non-responders (GAD-7: p=0.254; QIDS-SR16: p=0.508; Mann-Whitney U test; online supplemental figure 2C,D), suggesting that pre-existing anxiety or depressive symptoms did not account for the between-group difference in fatigue response. As an exploratory observation, two patients with a baseline GAD-7 score ≥10 improved to below this threshold following anifrolumab initiation (online supplemental figure 2E). These findings suggest that anifrolumab may exert an early clinical effect on fatigue that cannot be explained simply by disease activity control, reduced glucocorticoid burden or pre-existing anxiety or depression.

Figure 1. 58% of patients treated with anifrolumab showed improvement in fatigue. FAS score at baseline, 1 month and 3 months after anifrolumab initiation in (A) entire cohort (n=12), (B) fatigue responders (n=7) and (C) fatigue non-responders (n=5) are shown. SLEDAI at baseline, 1 month and 3 months after anifrolumab initiation in (D) entire cohort (n=12), (E) fatigue responders (n=7) and (F) fatigue non-responders (n=5) are shown. In (A) and (D), red lines indicate fatigue responders and blue lines indicate non-responders; different line styles distinguish individual patients. In (B), (C), (E) and (F), each symbol shape identifies an individual patient. (G) Scatter plots showing the relationships between PSL dose (mg/day), FAS score and SLEDAI in fatigue responders (n=7) at baseline, 1 month and 3 months after anifrolumab treatment. PSL dose versus FAS score (left panel), SLEDAI versus FAS score (centre panel) and PSL dose versus SLEDAI (right panel) are shown. Symbol shape identifies individual patients; fill shade indicates time point (dark grey: baseline; medium grey: 1 month; light grey: 3 months). Dashed lines connect sequential time points for each patient. Where values were identical at consecutive time points, the later time point is represented by a smaller symbol of the same shape superimposed at the same position. FAS, Fatigue Assessment Scale; PSL, Prednisolone; SLEDAI, Systemic Lupus Erythematosus Disease Activity Index.

Serum IFNα levels do not predict the anifrolumab response in fatigue

To assess whether elevated IFN responses predict anifrolumab fatigue response, serum IFN-α was measured by ELISA and compared between groups. The IFN-α concentrations were low in most patients and did not differ significantly between the groups (figure 2A). IFN-α/β reporter HEK293 cells to measure ISRE activation showed a similar trend, with comparable activity between the groups (figure 2B). THP1-dual cells also demonstrated comparable IFN-inducing activity between the groups (figure 2C). On the other hand, although not statistically significant, a trend towards higher NF-κB-activating activity was observed in a subset of responders as measured by THP1-dual cells (figure 2D). In addition to the IFN-α/β receptor and downstream signalling, THP-1 cells are known to respond to cell-free nucleic acids via the cGAS-STING pathway⁴⁴ and TLR pathway,⁴⁵ which may contribute to both IFN and NF-κB reporter signals. Z-score normalised values displayed as a heatmap (figure 2E) revealed that THP-1 ISRE and NF-κB activities were not significantly correlated (Spearman ρ = −0.028, p=0.931; figure 2F), indicating that these assays capture largely distinct aspects of immune activation. Given this orthogonality, an overall response (OR) composite score was defined as the maximum of the two z-scores per subject to capture patients with elevated activity in either pathway. Responders showed a numerically higher OR composite score than non-responders, although this did not reach statistical significance (figure 2G). These results indicate that serum IFN-α levels do not significantly predict fatigue response to anifrolumab; however, the heterogeneous IFN and NF-κB activation pattern in responders suggests that low-grade cellular responses, potentially driven by cell-free nucleic acids, may correlate to fatigue in a subset of patients.

Figure 2. Serum IFN-I levels do not predict fatigue improvement after anifrolumab.(A) Baseline serum IFNα2 concentration (pre-anifrolumab), measured by ELISA (VeriKine-HS Human IFN-α All Subtype ELISA Kit), in fatigue responders (n=7) and non-responders (n=5). (B) Baseline serum IFN-I bioavailability measured using IFN-α/β reporter HEK293 cells (SEAP, OD620nm), stratified by fatigue response. (C) Baseline serum IFN-I inducing activity measured using THP1-dual reporter cells (ISRE-luciferase, RLU), stratified by fatigue response. (D) Baseline NF-κB activating activity measured using THP1-dual reporter cells (NF-κB-SEAP, OD:620 nm), stratified by fatigue response.(E) Heatmap of z-score normalised values for HEK IFN bioavailability, THP-1 IFN inducing activity and THP-1 NF-κB activity across all 12 patients. Within each group, patients are sorted by THP-1 IFN inducing activity in descending order. (F) Scatter plot of THP-1 ISRE-luc (IFN inducing activity, x-axis) against THP-1 NF-κB-SEAP (NF-κB activity, y-axis); Spearman rank correlation coefficient (ρ) and p value are shown. (G) Individual z-scores for THP-1 ISRE-luc and THP-1 NF-κB activity and their overall response composite score (maximum of the two z-scores per subject), stratified by fatigue response. Horizontal bars indicate group medians. Group comparisons (A–D, G) were performed using the Mann-Whitney U test. p<0.05 was considered statistically significant. IFN, interferon; ISRE, interferon-stimulated response element; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; SEAP, secreted embryonic alkaline phosphatase.

γH2AX levels in pDCs are elevated in responders and decreased after anifrolumab treatment

Previous studies have demonstrated that γH2AX levels, a marker of the DDR, are elevated in B cells from SLE patients compared with healthy controls, and that IFN-α exposure induces DDR in healthy B cells in vitro.³² Although not directly comparable to our cohort given differences in disease activity, time point and cellular component, secondary analysis of TULIP-1 and TULIP-2 whole-blood transcriptome dataset demonstrated that expression of DDR-related genes was decreased following anifrolumab treatment (online supplemental figure 3). Given these backgrounds supporting IFN-DDR axis, we first measured γH2AX levels in peripheral blood B cell subsets from SLE patients in our cohort; however, γH2AX levels were not significantly different between fatigue responders and non-responders in any B cell subset examined (online supplemental figure 4). Since myeloid cells express the highest levels of IFN-stimulated genes in peripheral blood from SLE patients¹² and are IFN-sensitive cell types, we subsequently focused our analysis on myeloid cell subsets. To investigate cell-type-specific responses to cellular stress in relation to fatigue, γH2AX MFI was measured in monocytes, cDCs and pDCs from responders (n=6), non-responders (n=4) and healthy controls (n=6; figure 3A–C). Of the three cell types assessed, pDC γH2AX was significantly elevated in responders compared with both non-responders (p=0.044) and healthy controls (p=0.007; Dunn’s test with Bonferroni correction; overall Kruskal–Wallis p=0.005; figure 3A–C), whereas no such differences were observed in monocytes or cDCs. In responders, pDC γH2AX showed a non-significant decline (Wilcoxon signed-rank test, p=0.156) 1 month after anifrolumab initiation (figure 3F), suggesting that chronic cellular stress specifically in pDCs marks patients whose fatigue is responsive to IFN blockade. Furthermore, Spearman rank correlation between continuous baseline pDC γH2AX MFI and ΔFAS (3 month change) was significant (Spearman ρ = −0.70, p=0.025, n=10; figure 3G). These findings suggest that baseline pDC γH2AX may serve as a candidate predictive biomarker for fatigue response to anifrolumab, warranting validation in a larger prospective cohort.

Figure 3. γH2AX MFI in peripheral blood myeloid cell subsets across groups, before and after anifrolumab treatment in responders, and correlation with fatigue response. (A–C) γH2AX MFI in monocytes (A), conventional dendritic cells (cDCs; B), and plasmacytoid dendritic cells (pDCs; C) from healthy controls (n=6) and SLE patients classified as fatigue responders (n=6) or non-responders (n=4) to anifrolumab. Data are presented as median with IQR; individual data points are overlaid. (D–F) Longitudinal changes in γH2AX MFI in monocytes (D), cDCs (E) and pDCs (F) in the responder group at baseline (pre) and 1 month after anifrolumab initiation (post). Lines connect values from the same patient. (G) Scatter plot showing the correlation between baseline pDC γH2AX MFI and change in FAS score (ΔFAS; 3 months − baseline) across all patients (n=10). Spearman’s rank correlation coefficient (ρ) and p value are shown. Group comparisons (A–C) were performed using the Kruskal–Wallis test followed by Dunn’s post-hoc test with Bonferroni correction. Pretreatment and post-treatment comparisons (D–F) were assessed using the Wilcoxon signed-rank test. Numeric p values are shown for all comparisons. FAS, Fatigue Assessment Scale; HC, healthy control; MFI, mean fluorescence intensity.

Discussion

Our preliminary findings suggest that anifrolumab may provide fatigue relief in SLE patients with low disease activity and low baseline serum IFN-α. The fatigue response was associated with dynamics of γH2AX in pDCs: elevated at baseline and showing a declining trend after treatment. These observations suggest that IFN inhibition may address subclinical cellular stress even in the absence of overt systemic manifestations captured by standard disease activity scores, extending the observations of TULIP-1 and TULIP-2 to patients with low disease activity.

The interaction between systemic inflammation and the central nervous system (CNS) is increasingly recognised as a key driver of such symptoms: pro-inflammatory cytokines that act on the CNS can underlie ‘sickness behaviour’—lethargy, fatigue and social withdrawal—as observed in viral infections and experimental administration of LPS,⁴⁶ and IFN-α induces depressive-like behaviour in mouse models.⁴⁷ Clinical evidence from other conditions supports this mechanism. Longitudinal imaging of patients with melanoma receiving IFNα2 administration showed fatigue correlated with increased glucose metabolism in the basal ganglia.⁴⁸ ISG expression in peripheral blood correlates with fatigue dynamics more strongly than other inflammatory markers in patients with breast cancer,⁴⁹ reinforcing IFN signalling as a primary driver of fatigue in disease contexts.

We identified pDCs as uniquely sensitive to IFN-driven cellular stress, with γH2AX reduced by anifrolumab. Although precise triggers for DDR in SLE have yet to be defined, UV exposure generates oxidative DNA lesions, including 8-oxoguanine, which escapes TREX1 degradation and activates cGAS-STING, amplifying IFN production.^{25 50}

The trend towards higher THP-1 NF-κB activity in fatigue responders may reflect the presence of serum nucleic acids and immune complexes in these SLE patients, which were taken up by THP-1 cells,^{45 51} activating TLR-mediated NF-κB signalling preferentially over the IFN pathway. Notably, anifrolumab has been shown in vitro to suppress not only IFN-α but also TNF-α and IL-6 production from pDCs stimulated with Cytosine-phosphate-guanine (CpG) oligonucleotides or immune complexes²³; the reduction in pDC γH2AX after treatment may therefore partly reflect this broader cytokine suppression rather than IFN blockade alone. Direct experimental validation of this mechanism was not obtained in the present study and remains a subject for future investigation. As a speculative hypothesis without direct experimental support in the present dataset, we propose that, in our cohort, local rather than circulating IFN responses may be more critically involved in mediating fatigue: pDCs may accumulate IFN-driven cellular stress reflected by γH2AX during differentiation or peripheral tissue exposure, then re-enter the circulation, potentially explaining our findings despite low serum IFN-α concentrations. This model requires prospective experimental validation.

Several limitations should be acknowledged. The small sample size (n=12) and absence of a matched control group limit statistical power and may introduce selection bias. The uniformly low serum IFN-α concentrations in this cohort may reflect limited assay sensitivity of conventional ELISA; single-molecule array (Simoa)-based assays, which offer detection limits approximately 1000-fold lower than conventional ELISA, may enable quantification of biologically relevant IFN-α concentrations in patients with clinically quiescent SLE and warrant use in future studies. Additionally, concomitant immunosuppressive therapy and sample handling variables may further suppress detectable circulating IFN-α levels.

The FAS minimum clinically important difference (≥4 points) was derived from a sarcoidosis cohort and has not been formally validated in SLE; future studies should incorporate SLE-validated instruments such as Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue. Given that γH2AX was tested across seven cell populations without correction for cell type, the pDC-specific finding requires replication in a larger cohort. In a proportion of patients, γH2AX MFI appears to increase in monocytes and cDCs following anifrolumab treatment. Although monocyte γH2AX has been reported to correlate with SLEDAI in SLE,²⁷ SLEDAI was uniformly low in our cohort (median 4, IQR 4–7), leaving little room for disease activity-driven DDR reduction. Further analysis of subset composition within monocytes and cDCs is also required. Whether pDC γH2AX has genuine predictive value remains unclear. Flow cytometry MFI values also vary across instruments and laboratories, so the specific threshold identified here may not be directly transferable; identifying a surrogate marker as an alternative to pDC γH2AX would be needed for clinical application.

In conclusion, these hypothesis-generating findings suggest that targeting the IFN pathway may alleviate debilitating fatigue in clinically stable SLE. Large-scale controlled studies incorporating accessible biomarkers will be essential to validate these observations and personalise therapeutic strategies.

Data availability statement

Data are available upon reasonable request.