Distinct molecular signature in relapsed patients with ANCA-associated vasculitis

Abstract

Objectives

The pathogenesis of relapses in patients with microscopic polyangiitis (MPA) and granulomatosis with polyangiitis (GPA) remains ill defined. We aimed to identify whether there are any differences in the molecular profile between patients with relapsing versus new-onset disease.

Methods

Whole blood RNA sequencing in 21 patients with MPA or GPA and 12 age-matched/sex-matched healthy controls was performed; 11 patients had active disease, either newly diagnosed (treatment-naïve, n=6) or relapsed (off-therapy, n=5) and 10 were in remission, either on (n=4) or off (n=6) maintenance therapy. Differential gene expression and functional enrichment analysis were conducted.

Results

Relapsing patients with MPA/GPA exhibited a distinct molecular profile compared with healthy controls, newly diagnosed patients or those in remission. Specifically, patients with a major relapse displayed a unique pattern of 41 neutrophil degranulation-associated genes, encoding neutrophil granular proteins, proteins mediating neutrophil adhesion and chemotaxis and proteins involved in myelopoiesis.

Conclusions

In this well-defined MPA/GPA cohort, we report for the first time a distinct relapse-associated transcriptional signature consistent with enhanced neutrophil activity. Therapies targeting neutrophil activation could represent a novel approach for preventing or treating relapses in MPA/GPA.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• The pathogenesis of antineutrophil cytoplasmic antibody-associated vasculitides involves a complex interplay between innate and adaptive immunity pathways that ultimately culminate in small-vessel wall damage and target-organ inflammation.

• Relapses are not uncommon in patients with microscopic polyangiitis (MPA) and granulomatosis with polyangiitis (GPA), occurring during or after discontinuation of maintenance therapy, but the molecular mechanisms underlying relapses remain ill-defined.

WHAT THIS STUDY ADDS

• Patients with relapsing MPA or GPA display a distinct transcriptional signature, marked by enhanced neutrophil activation, including degranulation, which differentiates them from patients with early, newly diagnosed and treatment-naïve disease.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings reiterate the role of neutrophils in the pathogenic mechanisms associated with relapses in MPA and GPA, supporting the potential to alter the disease course through targeted therapeutic approaches.

Introduction

Microscopic polyangiitis (MPA) and granulomatosis with polyangiitis (GPA) are two forms of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitides (AAVs), characterised by inflammation in various organs and auto-antibodies against intracellular neutrophilic proteins (such as myeloperoxidase (MPO) and proteinase 3 (PR3)).^{1 2} In genetically predisposed individuals exposed to environmental antigens, the pathogenesis of AAVs involves early neutrophil activation, neutrophil extracellular trap (NET) formation and a loss of tolerance against intracellular neutrophilic proteins leading to ANCA production, and thus to a complex interplay between innate and adaptive immunity with activation of humoral, cellular and complement pathways that ultimately culminate in small-vessel wall damage and target-organ inflammation.^{1 2}

Despite advances in the management of MPA and GPA with the introduction of B-cell depleting agents such as rituximab (RTX), relapses remain common (10%–50%), unpredictable and associated with worse prognosis.^3–5 These relapses can occur either during maintenance therapy or after its discontinuation.^3–5

Our group and other investigators have examined the gene expression profiles of peripheral blood from AAV patients, proposing pathogenic molecular pathways^{6 7} and prognostic biomarkers for treatment response.⁸ Moreover, recent high-throughput technologies including single-cell RNA sequencing (RNA-seq) have provided a deeper insight into the AAV immunopathogenesis, paving the way to personalised prediction and targeted therapeutic strategies.^{9 10}

However, the molecular basis of relapsing disease, especially after discontinuation of maintenance therapy compared with treatment-naïve, new-onset disease has not been examined so far. We aimed, by analysing the whole blood transcriptome from patients with active newly diagnosed (treatment-naïve) or relapsing (off-therapy) AAV, as well as from those in remission (on and off maintenance therapy) and healthy controls, to identify whether the relapsing disease is associated with a distinct transcriptional signature in patients with MPA/GPA.

Patients and methods

Patients

We obtained whole blood from patients with MPA or GPA and age-matched/sex-matched healthy controls. All patients were followed-up in the joined rheumatology-nephrology outpatient clinic of the General Hospital of Athens ‘Hippokration’ (February 2019 to July 2023) and fulfilled the 2022 American College of Rheumatology/European Alliance of Associations for Rheumatology (EULAR) classification criteria for MPA/GPA.^{11 12}

We included eligible patients with new MPA/GPA diagnosis (treatment-naïve) or major relapse (off maintenance therapy), as well as patients in remission (either on or off maintenance therapy, online supplemental table S1).

Definitions of clinical disease activity and data collection

Disease activity was assessed by the Birmingham Vasculitis Activity Score V.3 (BVASv3). ‘Remission’ was defined as a BVASv3 of 0 plus prednisolone-equivalent dose <7.5 mg/day and ‘relapse’ as any increase in BVASv3 >0, after achieving BVASv3=0, with major relapses involving ≥1 major/vital organ, a life-threatening manifestation or both, necessitating new induction therapy. ‘Refractory’ disease was defined as unchanged/increased signs and/or symptoms of active disease after a period of induction therapy, as long as other potential causes of persistent/worsened disease manifestations are ruled out.¹³

For each patient, we recorded: sex, age, disease duration, AAV clinical phenotype (MPA/GPA), ANCA serotype (MPO-ANCA/PR3-ANCA positive or negative), previous relapses, organ involvement, disease activity and treatment history. Patients were followed-up from sampling until last visit/contact or death and the disease course was recorded.

Sample collection and total RNA isolation

Whole blood samples were collected in PAXgene tubes and stored at −80°C until further use.^{14 15} Total RNA was extracted using the QIAGEN PAXgene Blood RNA (IVD) Kit and quantification was conducted using a NanoDrop spectrophotometer.

Library preparation and RNA sequencing

RNA-seq experiments were performed in the Greek Genome Center of the Biomedical Research Foundation of the Academy of Athens. RNA-seq libraries were prepared with the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina, with 1 μg of total RNA input. Library QC was performed with the Agilent bioanalyser DNA1000 kit and quantitation with the qubit HS spectrophotometric method. Approximately, at least 17 million 101 bp single-end reads were generated for each sample. Sequencing was performed in NovaSeq 6000 (Illumina, San Diego, California, USA).

RNA sequencing data and functional enrichment analysis

Quality control was conducted on the FASTQ raw data file for each sample using the FASTQC software. FASTQ files were aligned to the hg19 genome using HISAT2.¹⁶ Counts were defined using the htseq-count command of HTSeq package V.2.0.3¹⁷ using ‘intersection non-empty’ mode combined with ‘nonunique none’ parameter and the option ‘reverse’ regarding the library strandness, as defined by the NEBNext Ultra II Directional RNA Library Prep Kit guidelines. Reads overlapping multiple features were discarded. The count files were used as input for DESeq2.¹⁸ Normalisation was performed with the estimate size factor function followed by differential gene expression analysis (DGEA), adjusting for age and sex. For the ‘relapsed versus newly diagnosed’ comparison, we also adjusted for ANCA serotype (MPO/PR3) and disease phenotype (MPA/GPA). Principal component analysis (PCA) after normalisation with the DESeq2 function ‘rlog’ was conducted to project the samples in dimensionality reduction plots. Differentially expressed genes (DEGs) were considered significant with |log₂ fold change|>1.5 and Benjamini-Hochberg adjusted p value <0.05 for multiple comparisons testing. Functional enrichment analysis was performed using the Enrichr web tool, with pathway databases (Reactome) and Gene Ontology terms (biological process).¹⁹ Finally, for the ‘relapsed versus newly diagnosed’ comparison, gene set enrichment analysis (GSEA) was also performed using all the genes ranked by the log₂ fold change.

Cell-type deconvolution of whole-blood RNA-seq data

The relative proportions of immune cell types were estimated using CIBERSORTx with the LM22 leucocyte gene signature matrix, which profiles 22 distinct immune cell subsets. Bulk RNA-seq count data from the patient cohort were normalised using the estimate Size Factors function from the DESeq2 package in R. The normalised expression matrix was analysed using the CIBERSORT R module, with 1000 permutations to estimate immune cell fractions. Samples with deconvolution p <0.05 were considered to have reliable results and were included in downstream analyses.²⁰

Statistical analysis

Continuous variables are presented as a median and interquartile range (IQR) and categorical variables as an absolute number and relative frequency (%), when required. Between-group differences were examined using the Mann-Whitney U test, while differences in immune cell proportions among the study subgroups were assessed with Kruskal-Wallis test in GraphPad Prism V.5.00 (GraphPad Software, USA).

Results

Patient characteristics

The patients’ characteristics according to their disease activity status are presented in table 1 and online supplemental table S2, and their individual disease and treatment trajectories at the time of blood sampling are depicted in figure 1. Overall, we included 21 patients with MPA/GPA (all ANCA-positive): 11 with active disease (newly diagnosed, treatment-naïve, n=6; relapsed and off therapy, n=5) and 10 in remission (on RTX maintenance therapy, n=4; off therapy, n=6), as well as 12 age-matched/sex-matched healthy controls.

Table 1. Patient characteristics.

	Active disease (n=11)		Remission (n=10)
	Newly diagnosed treatment-naïve N=6	Relapsed off therapy N=5	On maintenance therapy with RTX N=4	Off maintenance therapy N=6
Males, n (%)	2 (33)	4 (80)	2 (50)	3 (50)
Age at sampling (years), median (IQR)	61 (38–79)	70 (54–75)	71 (53–77)	68 (48–79)
Disease duration (years), median (IQR)	–	9 (4–10)	14 (5–20)	9 (5–13)
Previous relapse(s), n (%)	–	1 (20)	3 (75)	4 (67)
AAV clinical phenotype, n (%)
MPA	4 (67)	4 (80)	0	2 (33)
GPA	2 (33)	1 (20)	4 (100)	4 (67)
ANCA specificity (ever),* n (%)
MPO-ANCA	4 (67)	4 (80)	0	2 (33)
PR3-ANCA	2 (33)	1 (20)	4 (100)	4 (67)
BVASv3, median (IQR)	15 (11–21)	18 (10–21)	0	0
Organ/System involvement, n (%)			NA	NA
Constitutional symptoms	3 (50)	4 (80)
Lung	3 (50)	4 (80)
ENT	1 (17)	2 (40)
Kidneys	5 (83)	3 (60)
Skin	2 (33)	0
Mucus membranes/Eyes	0	1 (20)
Cardiovascular	0	2 (40)
Nervous	1 (17)	0
Other†	0	1 (20)
Time (years) from last treatment, median (IQR)	–	2.8 (2.5–4.3)	0.3 (0.3–0.4)‡	2.6 (0.7–4.3)
Last treatment, n (%)	–
RTX		1 (20)	4 (100)§	5 (83)
Cyclophosphamide		2 (40)	0	0
Methotrexate		2 (40)	0	1 (17)

At sampling, none was receiving glucocorticoids. Newly diagnosed patients were treatment-naïve. Relapsed patients were off therapy and had not yet received re-induction treatment.

^*Median (IQR) ANCA titre of newly diagnosed patients was 98 (70–125) IU/mL and of relapsed patients was 61 (50–93) IU/mL (with normal value <21 IU/mL) (Mann-Whitney U test: p=0.264).

^†Granulomatous prostatitis.

^‡Time (years) since the last RTX maintenance infusion.

^§All patients had previously received RTX for induction of remission and then continued RTX for maintenance of remission.

AAV, ANCA-associated vasculitis; ANCA, antineutrophil cytoplasmic antibody; BVASv3, Birmingham Vasculitis Activity Sscore V.3; ENT, ear-nose-throat; GPA, granulomatosis with polyangiitis; IQR, interquartile range; MPA, microscopic polyangiitis; MPO, myeloperoxidase; NA, non applicable; PR3, proteinase 3; RTX, rituximab.

Figure 1. Disease and treatment trajectories of all patients with MPA/GPA (n=21) up to the time of sampling. Patients are grouped according to their disease clinical status (newly diagnosed, relapsed, remission on RTX, remission off therapy) at the time of sampling. AZA, azathioprine; CYC, cyclophosphamide; GPA, granulomatosis with polyangiitis; MMF, mycophenolate mofetil; MPA, microscopic polyangiitis; MTX, methotrexate; RTX, rituximab.

In table 1, the clinical characteristics of patients with active newly diagnosed or relapsing disease are shown. No major differences were observed between these groups. Among newly diagnosed patients, 33% were males and 67% were MPO-ANCA+, with a median (IQR) age at sampling of 61 (38–79) years and a median (IQR) BVASv3 score of 15 (11–21); 83% had kidney and 50% lung involvement. Respectively, for relapsed patients, 80% were male and 80% were MPO-ANCA+, with a median (IQR) age of 70 (54–75) years and BVASv3 score of 18 (10–21); lung (80%) was the most frequent organ involvement, followed by kidney (60%), ear-nose-throat (40%) and cardiovascular (40%) manifestations. All relapsed patients (n=5) were off therapy for a median (IQR) time of 2.8 (2.5–4.3) years. They had previously received various therapies for induction (cyclophosphamide (CYC), n=3; methotrexate (MTX), n=2; RTX, n=1) and maintenance (MTX, n=2; RTX, n=1) of remission. For most of them (4/5), this was their first disease relapse (figure 1). Patients were followed up for a median (IQR) time of 2.0 (1.3–3.7) years after sampling.

Patients with active AAV versus healthy controls

In MPA and GPA, several pathways are involved in disease immunopathogenesis,²¹ but whether there is a distinctive molecular signature at disease onset versus relapse is unclear.

First, we investigated for differences between patients with active, newly diagnosed or relapsing disease and healthy controls. Through dimensional reduction plots in two-dimensional space, after unsupervised PCA, we observed in our cohort a better separation between ‘relapsed patients’ and ‘healthy controls’ than between ‘newly diagnosed patients’ and ‘healthy controls’ (figure 2A). DGEA in ‘newly diagnosed patients’ versus ‘healthy controls’ and in ‘relapsed patients’ versus ‘healthy controls’ revealed 269 and 375 DEGs, respectively (online supplemental tables S3, S4). In newly diagnosed patients, the upregulated genes (n=238) were mainly related to myeloid and erythrocyte differentiation (online supplemental figure S1), whereas in relapsed patients, the upregulated genes (n=335) were associated with neutrophil degranulation and innate immunity pathways (figure 2B, online supplemental figure S2).

Figure 2. Whole blood RNA sequencing in relapsed or newly diagnosed patients with MPA/GPA and healthy controls. (A) Principle component analysis for ‘relapsed patients and healthy controls’ and ‘newly diagnosed patients and healthy controls’. (B) Volcano plot depicting the DEGs for the ‘relapsed versus healthy controls’ comparison. Cut-offs for volcano plot: p-adjusted <0.05, |log₂ fold change| >1.5. (C) Venn diagram depicting the common and the exclusive upregulated DEGs between relapsed and newly diagnosed patients when compared with healthy controls. (D) Dot plot depicting the functional enrichment analysis for the upregulated DEGs found exclusively on relapsed patients (vs healthy controls) in Reactome database. (E) GSEA of relapsed versus newly diagnosed patients. Enrichment plot depicts the neutrophil degranulation pathway from Reactome. Genes are ranked according to their log₂(fold change) as defined from the differential expression analysis between relapsed and newly diagnosed patient groups. The running enrichment score (green line) reflects the degree to which genes in the pathway are over-represented at the top or bottom of the ranked list, with peaks indicating maximal enrichment. The NES and corresponding FDR q-value indicate the statistical significance of the pathway’s enrichment in relapsed samples compared with newly diagnosed samples. DEG, differentially expressed gene; FDR, false discovery rate; GPA, granulomatosis with polyangiitis; GSEA, gene set enrichment analysis; MPA, microscopic polyangiitis; NES, normalised enrichment score; PC, principal component.

Since the role of neutrophils is well established in the pathogenesis of AAVs,²² we also investigated whether an overlap exists in the upregulated genes of these two active disease subgroups (figure 2C, online supplemental table S5). Relapsed and newly diagnosed patients, compared with healthy controls, shared 114 (24.8%) common upregulated DEGs, which were associated with erythrocyte physiology, platelet activation and haemostasis (online supplemental figure S3). In line with the above results, the DEGs found exclusively upregulated in relapsed patients (vs healthy controls; n=221, 48.1%) were enriched in pathways associated with innate immunity, including neutrophil degranulation, type I interferon (IFN) response and activation of matrix metalloproteinases (figure 2D). On the other hand, the DEGs found exclusively upregulated in newly diagnosed patients (vs healthy controls; n=124, 27%) were associated with negative regulation of apoptosis, autophagy of mitochondrion and transforming growth factor (TGF)-beta receptor signalling (online supplemental figure S4).

Relapsed versus newly diagnosed patients

Next, to further elucidate the transcriptional differences between relapsed and newly diagnosed patients, we performed a head-to-head transcriptome comparison, yielding only 13 DEGs (online supplemental table S6). Despite the small number, we observed that the majority of these upregulated DEGs (9/12; namely DEFA4, LCN2, CTSG, AZU1, RNASE3, FOLR3, CEACAM8, ELANE and LTF) were involved in the neutrophil degranulation pathway (OR 93.64; online supplemental figure S5). Subsequently, to avoid potential false-negative results obtained by the ‘relapsed versus newly diagnosed’ comparison, given that they both represent active disease subgroups, we also employed GSEA preranked analysis to investigate whether neutrophil-associated pathways are involved at the gene signature of this comparison, ranking the DEGs according to the log₂ fold change parameter. Remarkably, we observed that the neutrophil degranulation pathway was also enriched at the unbiased gene signature (figure 2E).

Relapsed patients versus patients in remission

We have previously shown that patients with active AAV display an enhanced neutrophil-activation gene signature, compared with patients in remission.⁶ To comprehensively investigate the transcriptional aberrations in relapsed MPA/GPA, we compared the blood transcriptomes of relapsed patients (n=5) with those in remission either on (n=4) or off (n=6) maintenance therapy, separately.

PCA highlighted a clear separation of relapsed patients from patients in remission, regardless of therapy (online supplemental figure S6). Functional enrichment analysis of the upregulated DEGs in the ‘relapsed versus remission on RTX maintenance’ (n=16) and ‘relapsed versus remission off therapy’ (n=114) comparisons uncovered pathways relevant to neutrophil degranulation and Β-cell receptor signalling in the former, and neutrophil degranulation, translation and mitochondria-related metabolic processes in the latter comparison (online supplemental figure S7).

A neutrophil-associated transcriptional signature distinguishes relapsing from new-onset or remitting AAV

Based on the observed upregulation of neutrophil signature in relapsing MPA/GPA, we identified all the genes associated with the neutrophil-linked pathway terms derived from our functional enrichment analyses, and we defined the expression pattern of all neutrophil-associated DEGs that emerged from our analyses, among all participants. Remarkably, we found a unique upregulation of 41 neutrophil degranulation-associated genes, including, among others, AZU1, ELANE, DEFA4, LTF, BPI, CTSG, CAMP, SLPI, CEACAM6, CEACAM8, S100A9, S100A12, TARM1, MMP9, RNASE3, TNFAIP6 and PGLYRP1 (figure 3), which clearly distinguished relapsing patients from patients at diagnosis, in remission and healthy controls.

Figure 3. Heatmap depicting the gene expression levels (z-scores) of the neutrophil degranulation-associated genes across patients with MPA/GPA with active disease (newly diagnosed or relapsed) or in remission (on RTX or off therapy) and healthy controls. Red indicates higher and blue indicates lower expression relative to the mean expression of each gene across samples. Relapsed patients depict an overt upregulation at the expression levels of the depicted genes. DEG, differentially expressed gene; GPA, granulomatosis with polyangiitis; MPA, microscopic polyangiitis; RTX, rituximab.

Among newly diagnosed patients, one patient (#4) exhibited an upregulated expression profile of neutrophil degranulation-related genes (figure 3). Notably, this patient following first induction therapy with RTX had a refractory course with new lung lesions necessitating a second RTX induction. Also, during follow-up while on maintenance therapy with RTX and MTX, this patient experienced a major disease relapse with lung involvement (online supplemental table S7). Whether excessive neutrophil activation from disease onset could be associated with a subsequent refractory or relapsing disease course remains to be elucidated.

Finally, we performed a deconvolution analysis to determine whether our findings could be attributed to variations in the proportions of circulating immune cells among study subgroups. Hence, we observed that neutrophils represented the most abundant immune cell type across samples, followed by monocytes (figure 4A). However, no significant differences in neutrophil proportions were found between relapsed and newly diagnosed patients (post hoc pairwise adjusted p >0.999, figure 4B). Of note, no significant difference was observed in white blood cell differential counts between newly diagnosed and relapsed patients (online supplemental table S2). Thus, the observed upregulation of the neutrophil-related gene signature exclusively in relapsed patients could reflect an increased activation or functional reprogramming of neutrophils, rather than a shift in their abundance.

Figure 4. Cell-type Identification By Estimating Relative Subsets of RNA Transcripts eXtended (CIBERSORTx) analysis of immune cell composition in PBMCs from individual patient and healthy controls. (A) Stacked bar graph depicting the estimated proportion of immune cell types for every individual. Each bar represents the proportion of immune cell subsets within an individual sample, with different colours corresponding to specific cell populations as indicated in the legend. (B) Bar graphs depicting the estimated proportion of neutrophils (left panel) and monocytes (right panel) among the disease status groups of this study. The Kruskal-Wallis test was used to detect statistical differences among the groups, followed by Dunn’s multiple comparisons test. *P<0.05. NK, natural killer; ns, non-significant; PBMC, peripheral blood mononuclear cell; RTX, rituximab.

Discussion

Although the introduction of RTX as a first-line maintenance agent for patients with MPA and GPA has decreased the relapse rate compared with older agents like azathioprine,¹³ relapses are not uncommon, occurring either during⁴ or after discontinuation of maintenance therapy.²² However, the molecular and immunological mechanisms that underlie these relapses remain ill-defined. In our well-defined MPA/GPA cohort, whole blood transcriptome analyses revealed, for the first time, a distinct relapse-associated signature characterised by enhanced neutrophil activation in patients with relapsing disease.

Neutrophils play a pivotal role in the pathogenesis of AAVs. Specifically, a pathogenic vicious cycle involving ANCA production, neutrophil priming and activation followed by a pro-inflammatory response with respiratory burst, degranulation and NETosis amplifies and perpetuates the inflammation in the vasculature of target organs.^{7 21 23 24} Whether or not, there are potential differences in neutrophil activation patterns between new-onset or relapsing AAV, has not been studied so far.

In our study, patients with MPA/GPA who presented with a major relapse following a remission period off maintenance therapy, exhibited a distinct gene expression pattern associated with neutrophil degranulation that clearly distinguished them from newly diagnosed patients and those in remission (on RTX or off maintenance therapy). This gene set included, among others, genes encoding neutrophil granular proteins (AZU1, ELANE, DEFA4, LTF, BPI, CTSG, CAMP, SLPI and MMP9),⁸ cell surface proteins that mediate neutrophil adhesion (CEACAM6 and CEACAM8)²⁵ and genes involved in myelopoiesis and neutrophil chemotaxis (S100A8, S100A9, S100A12)²⁶ or toll-like receptor-mediated production of pro-inflammatory cytokines by neutrophils (TARM1).²⁷

Specifically, we identified exclusively in relapsed patients upregulated DEGs associated with neutrophil degranulation and innate immunity, including several genes that are highly expressed during the early maturation stages of neutrophils, such as LTF, DEFA3, LCN2, CAMP, S100A8, S100A12, MMP9 and SLPI.¹⁰ Indeed, neutrophils are currently considered a heterogeneous population, divided into subgroups according to their maturation status, surface markers and functions.²⁸ A recent single-cell transcriptome and surface proteome analysis of circulating neutrophils from patients with newly diagnosed MPA identified various neutrophil subsets with distinct functional characteristics, predominantly immature and primed neutrophils characterised by IFN-gamma/tumour necrosis factor (TNF) while expressing a type II IFN signature.¹⁰ Although these newly diagnosed patients exhibited a significant increase in a neutrophil subset with an immature phenotype, enriched in neutrophil degranulation-related genes, these genes were not differentially expressed in newly diagnosed patients compared with controls, in line with our findings.

Moreover, relapsed patients compared with healthy controls, but not newly diagnosed patients, exhibited upregulation of type I IFN-regulated genes, including MX1, IFI6, ISG15, IFIT1, UBA52, IFIT3 and OASL. We have previously observed that the type I IFN signalling represents a robust signature particularly in MPA,²⁹ while emerging data highlight its association with disease prognosis and kidney pathology in patients with MPO-AAV.³⁰ Hence, the role of type I IFN in AAVs appears to be critical but remains to be further elucidated.³¹

Furthermore, relapsed patients compared with patients in remission (both off therapy) displayed a transcription signature of increased mitochondria-related metabolic processes. Metabolically, neutrophils mainly rely on glycolysis, while oxidative phosphorylation has a limited role due to the low mitochondrial density. However, subsets of immature neutrophils have been reported to engage in oxidative mitochondrial metabolism.³² Indeed, we have previously identified a molecular endotype of active MPA/GPA that was associated with deregulation of metabolic profile and specifically mitochondrial dysfunction, being accompanied by neutrophil activation.⁶

Taken together, our findings suggest that relapses in MPA and GPA are characterised by pronounced neutrophil-driven inflammation, possibly dominated by immature neutrophil subsets. Although neutrophils have traditionally been considered as rapid responders of the innate immune system, lacking memory due to their short lifespan, recent evidence supports the concept of ‘trained immunity’ extending beyond monocytes to neutrophils. Interestingly, this process is thought to occur early in bone marrow through epigenetic and metabolic reprogramming of neutrophil precursors.^{33 34} More targeted studies are needed to examine the presence and significance of trained immunity in neutrophils within the context of AAVs.

In contrast, patients with newly diagnosed MPA/GPA compared with healthy controls were exclusively characterised by a transcriptional signature with enhanced TGF-beta signalling, negative regulation of apoptosis and enhanced mitophagy. TGF-beta signalling stimulates various cellular responses for immune homeostasis, while its dysregulation has been associated with autoimmune diseases, including AAVs.³⁵ Mitophagy represents a selective autophagic mechanism for the clearance of redundant or defective mitochondria as to maintain quality control. Otherwise, mitochondria can constitute sources of damage-associated molecular patterns.³⁶ Indeed, in patients with active MPA/GPA, increased levels of circulating mitochondrial DNA have been reported compared with healthy controls, which may contribute to inflammatory responses in AAVs.³⁷ Hence, this gene signature may reflect the cellular stress response signals and the metabolic reprogramming that predominate during the early stages of AAV course.

Interestingly, we observed that the only patient with newly diagnosed disease who was refractory to induction therapy and subsequently relapsed while on maintenance therapy (online supplemental table S7, newly diagnosed #4), exhibited an overt upregulation of neutrophil-related genes. A previous RNA-seq study in whole blood from AAV patients in the Rituximab in ANCA-Associated Vasculitis (RAVE) trial, which included both newly diagnosed and relapsed patients, revealed higher expression of neutrophil granular proteins (ie, MPO, PR3, ELANE, BPI, AZU1, CTSG, LTF, DEFA3, DEFA4, LCN2 and CAMP) in patients who did not respond to induction therapy with RTX or CYC/azathioprine compared with responders.⁸ Thus, an enhanced neutrophil-driven inflammation at disease onset may also indicate a refractory or relapsing disease phenotype. Obviously, longitudinal studies with larger number of patients and detailed trajectory analyses are needed to clarify temporal relationships and potential predictive value.

A major strength of our study is that we used well-defined disease subgroups according to their clinical status and treatment. To the best of our knowledge, this is the first study comparing the molecular profile of AAV patients with active disease between onset and relapse. Furthermore, none of the patients with active newly diagnosed or relapsing disease were receiving any type of immunomodulatory therapy at the time of sampling, including glucocorticoids, which are known to recruit neutrophils from the bone marrow and affect their function.^{38 39} Remarkably, the deconvolution analysis further supported our findings, revealing no significant differences in the proportion of circulating neutrophils between newly diagnosed and relapsed patients.

However, certain limitations should be considered. First, given the rarity of these diseases, we had a relatively small number of participants in each subgroup, while longitudinal samples were not available to examine within patient transcriptional changes across the disease course. Thus, we instead performed a cross-sectional comparison of newly diagnosed versus other disease states and especially relapsing disease. It is obvious that this is a hypothesis-generating study and its findings need to be validated in other cohorts. Second, most patients with active disease were MPO-ANCA-positive and thus we adjusted our analysis to reduce potential confounding. However, residual confounding cannot be excluded and so the observed transcriptional differences could reflect mechanisms driving relapses, and could represent downstream consequences of active inflammation during relapse, underlying patient characteristics, including differences in treatment history, or relapse-prone phenotypes. Hence, a larger group of patients with relapsing AAV is required to investigate potential differences in neutrophil-related responses between MPO-ANCA-positive and PR3-ANCA-positive disease. Third, we performed whole blood bulk RNA-seq, losing critical information regarding cell type and tissue gene expression. However, we used strict statistical criteria, with our findings demonstrating consistently in all comparisons an increased expression of neutrophil-related genes in relapsed patients.

In conclusion, this is a discovery study, which for the first time provides robust evidence that the molecular signature of active disease in MPA and GPA differs by disease stage. Patients with major relapses exhibited a unique transcriptional profile, different from newly diagnosed patients, marked by a gene panel specific for neutrophil activation. In addition, exclusively relapsed patients compared with healthy controls, but not newly diagnosed patients, exhibited upregulation of type I IFN-regulated genes. These findings offer novel insights into the role of neutrophils in the pathogenic processes that may precede or accompany relapses in AAVs, supporting the potential to alter the disease course through targeted therapeutic approaches. Obviously longitudinal as well as further studies are needed in order to confirm these findings.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.