The American College of Rheumatology text on VEXAS: guidance, not a recommendation

Why this text exists

VEXAS syndrome was individualised in 2020 and more than 450 cases have been published since. When the panel was formed, no clinical document addressed the physicians who care for these patients: this is what the Methods section of the article gives as its starting point, and it is the reason for the work.

The American College of Rheumatology describes its guidance documents as a route open to projects that seek its endorsement or its collaboration, that do not fit its existing publication channels, and for which the rheumatology community is asking for clinical direction. The definition says exactly what is at stake here: a disease too recent to carry an evidence-based recommendation, and a demand from the field that is not waiting for the evidence to exist.

The text was submitted on 13 December 2024, accepted on 25 April 2025, published online on 11 August 2025 and paginated in the March 2026 issue of Arthritis & Rheumatology. The founding workshop was supported by the FAI²R rare disease network; the text carries the endorsement of EuroBloodNet and of the ISSAID society.

Guidance and recommendation: where the difference lies

The distinction is not a matter of vocabulary. It can be read in the method, and six points establish it, all of them written in the article.

No formal systematic review. The subgroups carried out a review of the literature from January 2020 to August 2024 on PubMed and Medline, covering articles in English and abstracts from the international rheumatology and haematology congresses. The article states that no formal systematic review was conducted by the subgroups, the disease being too recent and the data too scarce.

Every statement is conditional. The authors write this, and they give the reason: the absence of any randomised controlled trial in VEXAS. The GRADE method was used to rate the quality of the evidence supporting each statement; it did not produce a gradation of strength between strong and conditional statements.

A single voting round. After the two-day workshop held in Paris in May 2024, voting was electronic and anonymous, yes or no, with an obligation to give reasons for disagreement, and a consensus threshold set at 80 % agreement. There was no second round. Every statement put forward crossed the threshold.

An expert panel, and no one else. Fifty-seven members, eighteen countries, nine disciplines: dermatology, genetics, haematology, immunology, internal medicine, rheumatology, infectious diseases, intensive care and pathology. The article states that patient representatives were not included. For its clinical practice guidelines, the American College of Rheumatology on the contrary convenes a patient panel, which gives its opinion just before the voting panel meets, and seats at least two of its members on the voting panel itself.

The authors themselves say what is missing. Their conclusion states that evidence-based recommendations, developed with the participation of patient partners, will remain to be produced as data accumulate. The text therefore presents itself as a stage, not as a point of arrival.

The bibliographic databases, for their part, index it as a guideline. The MEDLINE record for the article carries the publication type Practice Guideline. A search engine, a reference manager or a monitoring tool will therefore present it as a practice guideline. This is neither what the American College of Rheumatology says, nor what the authors write. The gap deserves to be known by anyone citing this text in a file or in a decision.

What this implies in practice: a conditional statement describes a course of action that the panel considers reasonable given the current data, it does not carry the weight of a clinical practice guideline and it does not remove the need for an individual decision. Agreement of 100 % between experts measures their agreement, not the level of evidence. The authors themselves rate as low the quality of the data supporting the treatment considerations, for reasons they name: selection bias, insufficient follow-up duration and insufficient precision in the diagnosis of associated myelodysplastic syndrome, and the absence of a standardised definition of treatment response.

Twenty-eight statements, and what their spread indicates

The statements fall into four domains: clinical and laboratory features (six), testing for the UBA1 mutation (seven), diagnosis of a myelodysplastic syndrome in a VEXAS patient (four), and outcome, prognostic factors and management (eleven). The total of twenty-eight is written nowhere in the article: it is obtained by enumerating table 1, and the count was redone for this page. Agreement rates run from 87 % to 100 %, with fifty-three voters on the first domain and fifty-two on the other three.

Two observations, drawn from the table itself. Six statements are unanimous, and they are the most general: the disease is not inherited, bone marrow vacuoles are suggestive without being diagnostic, the majority of typical forms involve exon 3, management must be multidisciplinary, glucocorticoids are usually necessary, and trials will be necessary. Conversely, the three least consensual statements are the most operational: the effect of ongoing treatment on the reliability of the genetic test (87 %, the lowest agreement of the twenty-eight), what to do when a blood test is negative despite strong suspicion (90 %), and the definition of a flare (90 %). Agreement narrows where the decision is made.

When to request the test, and how to request it

The panel holds that a persistent elevation of inflammatory markers, associated with skin, eye, lung or cartilage manifestations, or with a cytopenia, should raise the possibility of VEXAS. The pre-test probability is higher in men, after the age of 50, and when acute phase proteins are raised. Macrocytic anaemia is the most frequent haematological abnormality without being constant; macrocytosis may be isolated and may precede the rest. Vacuoles in erythroid or myeloid precursors are suggestive without being diagnostic: they are also found in deficiency states, copper deficiency in particular, and after chemotherapy. Their absence should not lead to abandoning the test when suspicion persists.

Seven statements concern testing for the mutation, and four points change the way a test request is written.

1. The majority of typical presentations involve a missense or splice site variant in exon 3 of UBA1, around methionine 41. This is the statement voted unanimously by the fifty-two voters. Sanger sequencing of exon 3 also covers the Ser56 residue, whose variants give a more moderate inflammatory phenotype.
2. The choice of technique sets the detection threshold. The text states that Sanger sequencing detects little below a variant allele burden of about 20 %, whereas targeted next-generation sequencing, droplet digital PCR and high-resolution melting analysis reach down to around 1 %. A test targeted on exon 3 remains an option, provided one knows that it misses variants located outside exon 3.
3. Ongoing treatment lowers the mutational burden and may distort the result, in particular under a hypomethylating agent. This is the least consensual statement, and it is also the one with the greatest consequences: the timing of sampling relative to treatment is part of the request.
4. When the blood test is negative and suspicion persists, the panel indicates that one should check that the analysis did cover the whole gene, then consider a bone marrow sample, using the method most sensitive to low variant allele burdens. The mutation is myeloid and absent from lymphocytes; it has been detected in skin or nails, but possibly through blood contamination, since fibroblast culture does not find it.

One figure that is useful in the clinic: the combination of bone marrow vacuoles and relevant clinical signs gave a positive predictive value of 87.5 % for a pathogenic UBA1 variant, in the study the text cites on this point. Genetic testing remains necessary for diagnosis in all cases.

The haematological side, and the trap of diagnosing myelodysplastic syndrome

Four statements. Bone marrow aspiration is indicated in a VEXAS patient with a cytopenia, in order to rule out an associated haematological disease, and the text recommends it before starting a disease-modifying treatment. It is accompanied by a karyotype and by next-generation sequencing looking for coexisting somatic mutations. When the diagnosis has been made by Sanger sequencing or by a test targeted on UBA1, the text indicates that the myeloid next-generation sequencing panel should be performed nonetheless. Testing for loss of the X chromosome has particular value in women.

Interpreting the marrow is the difficult point. Signs of dysplasia are frequent without the WHO and ICC 2022 criteria for myelodysplastic syndrome being met, and vacuolisation alone is not enough to conclude to dysplasia, since it belongs to the disease itself. The text reports that the literature places the frequency of associated myelodysplastic syndrome between a third and a half of patients, but that with a stricter definition, requiring transfusion-dependent anaemia and a platelet count below 100 G/L in addition to morphological or genetic abnormalities, this frequency falls to 18 %. Depending on the definition used, the same patient either is or is not considered to have a myelodysplastic syndrome.

The text settles one frequent situation: faced with isolated vacuoles, without other dysplastic features as defined by the WHO, with a normal karyotype and no other somatic mutation, the consensus is not to retain the diagnosis of myelodysplastic syndrome and to tighten monitoring. It also settles the other direction: a blast count of 5 to 20 %, which is rare, or a cytogenetic abnormality defining a myelodysplastic syndrome do lead to that diagnosis. The clonal landscape is dominated by DNMT3A and TET2, the karyotype is normal in more than half of cases, and mutations in myeloid driver genes are present in close to 60 % of patients, with or without myelodysplastic syndrome.

Treatment: what the panel ranks, what it rules out

Eleven statements concern outcome, prognosis and treatment. Median survival from symptom onset is about ten years, according to the figure the text takes up.

Glucocorticoids. They are usually necessary, at moderate to high doses, the article citing the order of 10 to 25 mg of prednisone per day or more, with slow tapering to the lowest effective dose. The text specifies that glucocorticoid-free remission is unusual in VEXAS, unlike in many rheumatological diseases, and that small dose reductions are sometimes enough for inflammation to start again.

Treatments targeting inflammatory pathways. Janus kinase inhibitors and interleukin 6 inhibitors are presented as more effective than conventional disease-modifying treatments, methotrexate and azathioprine, and than treatments directed against B cells. Interleukin 6 inhibitors are the most widely used. Among Janus kinase inhibitors, ruxolitinib may be preferred, in particular because it can be titrated over a wider dose range, the article giving 10 mg twice daily to 20 mg twice daily. Monitoring covers the worsening of cytopenias, and patient selection takes account of the increased risk of viral infection, thromboembolic disease and cardiovascular events specific to this class, with anti-alphaherpesvirus prophylaxis to be considered. Interleukin 1 inhibitors may be considered, with the reservation of the severe injection site reactions reported with anakinra and the limited data on canakinumab.

What the text rules out. B cell depleting treatments are said to be contraindicated: they are not effective in VEXAS and may worsen the B cell lymphopenia associated with the disease. This is a statement that can be acted on directly by a clinician receiving a patient already treated for another diagnosis.

Treatments reducing the clone. Azacitidine may be effective in some patients, with a decrease or even a disappearance of the UBA1 mutational burden in blood or marrow and improvement of inflammation. Treatment discontinuations with prolonged remission have been reported in small observations, but long-term data and prospective comparisons are lacking, and the text writes that it is not yet known whether patients with an associated myelodysplastic syndrome are the best candidates or whether the benefit extends more widely.

Allogeneic transplantation. It may be curative, the mutation being restricted to the haematopoietic compartment, and should be reserved for selected patients after careful haematological assessment. Candidates are patients with progressive and severe bone marrow failure, in particular transfusion-dependent, or with refractory inflammation. Thirty-three transplanted patients had been reported at the time of writing, of whom twenty-seven were alive, that is 81.8 %, and eleven had post-transplant molecular data showing complete eradication of the UBA1 mutation, most often over a follow-up of less than twelve months. The text designates the timing of transplantation and the selection of candidates as the most difficult question, because of age, comorbidities and the general condition impaired by the disease and by glucocorticoid therapy.

Three preventive measures to consider in every patient

A single statement, voted at 98 %, brings together three preventive measures: prophylaxis of opportunistic infections, in particular pneumocystosis and alphaherpesvirus infections, with particular attention to elderly patients and to those with a severe cytopenia on immunosuppressants; prevention of thromboembolic disease, present in about half of patients, mostly venous but also arterial, with thromboprophylaxis in at-risk situations such as hospitalisation, unless contraindicated; and limitation of the adverse effects of prolonged glucocorticoid therapy, with attention to bone and to cardiovascular risk.

Two practical points are added to these, written in the body of the text. When a skin lesion does not respond to treatment, a biopsy should be considered to rule out a non-tuberculous mycobacterial infection before increasing immunosuppression. And when any new or recurrent symptom appears, the differential diagnosis must remain broad: infections, thromboembolic disease and treatment complications are frequent and can mimic a flare. C-reactive protein is a more reliable marker of inflammation than the erythrocyte sedimentation rate in this disease, since haematological abnormalities raise the latter independently of inflammation.

What the text does not settle

The last statement, voted unanimously, indicates that clinical trials will be necessary to determine which treatment is the most effective. This is not a closing formula: it is the exact description of the state of the data.

• No randomised controlled trial had been conducted at the time of writing. Most of the published data are isolated observations or small series, and prospective cohorts are lacking.
• The efficacy of no drug is established beyond glucocorticoids, in the authors' terms. The hierarchy between classes rests on retrospective series, and is therefore exposed to indication bias.
• The timing of starting treatment is not settled, even though the disease is a progressive clonal process of the marrow: the authors indicate that this timing could weigh as much as the choice of drug.
• The duration and choice of anticoagulation after a thromboembolic event are not defined.
• The candidates for hypomethylating agents are not identified.
• The timing of allogeneic transplantation and the selection of candidates are designated as the most difficult question.
• At the date of the text there was neither a standardised definition of treatment response nor a validated activity score. This is one of the reasons why the authors rate the quality of the data as low.
• The true frequency of the disease is not known. The text takes up a prevalence of pathogenic UBA1 variants of 1 in 13,591 individuals, measured in a single American health system, and draws from it in its introduction an order of magnitude close to one million people worldwide, while writing that complete global data are lacking. This extrapolation rests on that study alone.

What has moved since the literature search closed

The literature review stops in August 2024. Three works published in 2026 bear on the points the text left open: a consensus definition of flare for clinical research, a specific activity index, the VEXAS-DAI, and the protocol of the PAXIS trial, which assesses pacritinib against placebo. The first two answer the absence of standardised definitions; their prospective validation remains to be done. The third is a protocol article: it describes the design of a trial, it reports no results. Arsène Mekinian is a co-author of all three.

His part in this text

Arsène Mekinian is the first of the fifty-eight named authors. The article designates eight co-first authors who contributed equally, of whom he is one, and five co-last authors. He is one of the two corresponding authors, with the last named author, David B. Beck, the correspondence address line of the article naming both. The affiliation carried on the article is the department of internal medicine of Saint-Antoine Hospital, AP-HP, Sorbonne Université, with the Imagine Institute and the reference centre.

An indexing note, useful to anyone citing this work: the journal signs it « Arsene Mekinian », without an accent, including on the first name, whereas other databases index it with accents. The article also carries three correction notices: on 27 August 2025, on the affiliation of two authors; on 21 October 2025, stating that the names of Arsene Mekinian and Sophie Georgin-Lavialle were corrected after first online publication; and on 11 March 2026, on the name of Mael Heiblig. Citations produced before these dates may therefore carry incorrect spellings. The ORCID identifier attached to the signature is 0000-0003-2849-3049.

References

1. Mekinian A, Georgin-Lavialle S, Ferrada MA, et al. American College of Rheumatology Guidance Statement for Diagnosis and Management of VEXAS Developed by the International VEXAS Working Group Expert Panel. Arthritis Rheumatol. 2026;78(3):509-522. DOI 10.1002/art.43287. PMID 40787890. Published online on 11 August 2025. Open access, CC BY-NC-ND 4.0 licence. Free full text: PMC12991921. Unless stated otherwise, all the data on this page come from it.
2. American College of Rheumatology, guidance documents page: rheumatology.org/guidance-documents. The text is referenced there under the title Diagnosis and Management of VEXAS: Consensus Statements from an International Expert Panel, dated 11 August 2025. Page consulted on 10 September 2026.
3. American College of Rheumatology, clinical practice guidelines page: rheumatology.org/clinical-practice-guidelines. Patient Involvement section. Page consulted on 13 September 2026.
4. Weeks LD, et al. Establishing a consensus definition of VEXAS flare for clinical research. Rheumatology (Oxford) 2026;65(2):keaf624. PMID 41289141.
5. Byram K, et al. Development of a Disease Activity Index for the Assessment of VEXAS Syndrome (VEXAS-DAI). Arthritis Care Res (Hoboken) 2026. DOI 10.1002/acr.80117. PMID 42439317.
6. Beck DB, et al. PAXIS: a randomized, double-blind, placebo-controlled, dose-finding phase 2 study (part 1) followed by an open-label period (part 2) to assess the efficacy and safety of pacritinib in patients with VEXAS syndrome. J Clin Med 2026;15(4):1426. PMID 41753113.

The studies cited inside the American College of Rheumatology text (prevalence, predictive value of vacuoles, median survival, clonal landscape, allogeneic transplantation) are cited as that text reports them; their own records were not reopened for this page.

Where to find this publication

The page on this site is the original. The document discussed here is openly accessible on the publisher's site, under a CC BY-NC-ND 4.0 licence: no figure, no table and no extract from that document is reproduced here.

The full text can also be read free of charge on PubMed Central: PMC12991921. The press release published by the American College of Rheumatology on 11 August 2025 is online at rheumatology.org.

Disclosures

Statement required by article R.4113-110 of the French public health code. The text discussed on this page is signed by the author of this page, who is its first author and one of its two corresponding authors: that is the first interest to declare here. The declaration relating to the subject covered, including links with the companies that manufacture the drugs mentioned, is set out on the Disclosures page of this site.

Page written for professional information. It does not replace individual medical advice and does not constitute a treatment protocol.