JAK2 is a critical therapeutic target in VEXAS syndrome treated with ruxolitinib

Abstract

Vacuoles, E1 enzyme, X-linked, autoinflammatory and somatic (VEXAS) syndrome, first identified by Beck et al.,¹ is an acquired monogenic autoinflammatory disorder characterized by multisystemic relapsing inflammation, often accompanied by multilineage cytopenia, venous thromboembolism, myelodysplastic syndrome and monoclonal gammopathy. VEXAS is caused by post-zygotic mutations of the ubiquitin-like modifier activating enzyme 1 (UBA1) gene within haematopoietic stem cells, resulting in the disruption of the ubiquitin–proteasome pathway.^{1, 2} Recent research indicates that these aberrations ultimately cause abnormal myeloid cell cycling and increased cytokine production by monocytes.^{3, 4} Nevertheless, the pathophysiology of VEXAS remains only partially understood, preventing the identification of actionable therapeutic targets.

Consistently with these knowledge gaps, few therapeutic options are currently available for the management of VEXAS. Indeed, while high-dose steroids are beneficial during disease flares, their long-term utilization is discouraged by their side effects. On the other hand, conventional disease-modifying anti-rheumatic drugs are scarcely useful for the management of inflammatory symptoms.^{2, 5, 6} Recent retrospective studies have identified Janus kinase inhibitors (JAKi) as effective therapeutic agents for patients with VEXAS.⁶ Of these, ruxolitinib, a dual inhibitor of JAK1 and JAK2, is associated with higher response rates compared to other JAKi.⁶ However, the mechanisms by which patients with VEXAS respond more to ruxolitinib than other JAKi are still unclear. Addressing this knowledge gap is critical for discovering the specific actionable targets in VEXAS, possibly allowing for the optimization of current therapeutic strategies.

To dissect the molecular characteristics of response to dual JAK inhibition, we charted the transcriptomic response to therapy in one patient with VEXAS treated with ruxolitinib. The patient, a 63-year-old Caucasian man, was diagnosed with VEXAS by genetic testing in April 2023 following an extensive diagnostic work-up for unexplained inflammatory symptoms, empirically treated with corticosteroids. VEXAS coexisted with WHO-defined myelodysplastic syndrome with low blasts⁷ and immunoglobulin M monoclonal gammopathy of undetermined significance (MGUS). Azacytidine was initiated in mid-May 2023, with contemporary steroid tapering. By early June, the patient required hospitalization due to high remittent fever, diffuse purpuric rash and enanthem. The biopsy of a purpuric skin lesion revealed dermic neutrophil and lympho-monocytic infiltrates positive for myeloperoxidase, confirming the diagnosis of VEXAS flare. Following initial treatment with intravenous steroids, ruxolitinib was initiated as steroid-sparing agent, resulting in complete clinical remission within 2 weeks (Figure 1A). Ruxolitinib was later discontinued due to worsening pancytopenia, necessitating red cell transfusions. Azacytidine was then resumed, showing good tolerance and a sustained response.

To comprehensively profile with high temporal resolution of the response to ruxolitinib, we performed serial blood RNA-sequencing (RNA-seq) coupled with bulk and single-cell RNA-seq (scRNA-seq) of VEXAS skin lesions (Figure 1A). Peripheral blood samples were obtained at multiple time points during treatment with ruxolitinib throughout patient's hospitalization and discharge. Specifically, blood samples were collected on day 0 before ruxolitinib initiation and on days 1, 3, 5, 6, and 11 after treatment started. Formalin-fixed, paraffin-embedded skin biopsies obtained per clinical practice were collected before starting steroid therapy (2 days before ruxolitinib initiation, day −2) and on day 5 after ruxolitinib initiation for bulk RNA-seq. Additionally, a fresh skin tissue sample was collected on day 5 for scRNA-seq. A detailed description of the materials and methods, along with the ethics statement, is provided in the Supporting Information.

To verify whether blood RNA-seq was informative of disease activity, we first assessed a transcriptomic VEXAS-specific signature identified by Kosmider et al. and comprising IL-1α, IL-1β, IL-18, TGF-α, IL-7, LGALS3, S100A8 and S100A9.⁴ Treatment with ruxolitinib was associated with a prompt downregulation of this signature (Figure 1B), in parallel with the progressive resolution of symptoms. Indeed, the VEXAS signature positively correlated with several laboratory markers of inflammatory activity assessed per clinical practice, including red cell distribution width, neutrophil percentage and absolute count, C-reactive protein, ferritin and lactate dehydrogenase. Conversely, it was negatively associated with monocyte percentage and absolute count, mean corpuscular haemoglobin concentration, as well as lymphocyte percentage and absolute count (Figure 1C; Table S1).

To shed light on the molecular determinants of response to JAK inhibition, we next profiled the transcriptional activity of the JAK–signal transducer and activator of transcription proteins (STAT) signalling pathway, observing a progressive downregulation of the JAK–STAT signature following ruxolitinib initiation (Figure S1). Notably, when considering the individual transcripts of JAK1 and JAK2, we observed a marked downregulation of JAK2, while JAK1 expression did not change over time (Figure 1C). JAK2 was positively associated with the VEXAS signature, as well as with interferon signalling, a key inflammatory pathway downstream of JAK-STAT. JAK2 expression was also associated with plasma cell response, indicative of MGUS activity. JAK1 displayed an opposite trend compared to JAK2 (Figure 1D).

Considering bulk RNA-seq performed on VEXAS skin lesions, the transcriptomic changes occurring after treatment initiation recapitulated those observed in blood, with the downregulation of VEXAS and interferon signatures, coupled with JAK2, but not JAK1, transcript decrease (Figure 2A). Immune cell deconvolution revealed a concomitant shift from a pro-inflammatory environment to a tissue repair status, with the complete loss of neutrophils and monocytes and a switch of macrophages from an M1 pro-inflammatory phenotype towards an M2 anti-inflammatory one (Figure 2B), consistent with clinical improvement. scRNA-seq analysis performed on fresh skin tissue collected on day 5 further confirmed the differential expression of JAK1 and JAK2 genes after treatment initiation (Figure 2C). Notably, both genes were mainly expressed on clusters of fibroblasts, vascular endothelial cells and pericytes. Indeed, JAK-STAT signalling is known to modulate endothelial permeability, inducing a pro-adhesive and pro-coagulant switch in inflammatory conditions.⁸

Overall, our results indicate that JAK2 is a critical actionable target in patients with VEXAS, which has been suggested in clinical reports but never described from a molecular standpoint.⁶ Despite the efficacy of ruxolitinib, its discontinuation due to pancytopenia in the above-described case underscores the need for careful management of its haematological side effects. Our findings provide a strong pathophysiological basis to investigate the efficacy of selective anti-JAK2 molecules in patients with VEXAS, potentially reducing the rate of therapy-related side effects while preserving the efficacy in inducing and maintaining clinical remission.

Conceptualization: GZ, LF and AlB, methodology: GZ, LF and MD, data collection: NG, AnB, BC, MS, AC, MD, IL, TV, GR, MD, IS and AG, data analysis: LF, GZ, FR, MD, MC and AC, software: LF, formal analysis: AC, MD, FR, LF, AlB and GZ, writing and review: all authors, visualization: LF, GZ and FR, supervision: AlB, LF and GZ, project administration: MD and GZ and funding acquisition: AlB and GZ.

GZ declares consultancy fees from Menarini Stemline and holds the co-ownership of Immunomica Ltd. The remaining authors have no conflict of interest to disclose.