Methodology and Clinical Significance of Longitudinal UBA1 Clonal Burden Evaluation in Vexas Syndrome

VEXAS is a hemato-inflammatory syndrome genetically defined by somatic mutations in the X-linked UBA1 gene in hematopoietic precursor cells. Val/Thr/Leu substitutions at the Met41 hotspot constitute the molecular underpinnings of VEXAS in more than 90% of patients, typically men during the 6-7th decades of life. Clinical manifestations of the disease are extremely various and tend to relapse after most hematological and rheumatological treatments. Based on patients’ phenotypes, therapy can either target the UBA1 mutant clone (e.g., with azacitidine or allogeneic hematopoietic cell transplantation [allo-HCT]) or control the inflammation and hypercytokynemia germane to the syndrome (e.g., with DMARDs). To date, no guidelines exist for the management of VEXAS, and scarce is the evidence on methodology and clinical significance of longitudinal UBA1 clonal burden evaluation upon therapy. To address this gap in knowledge, we devised a new ddPCR method to quantify UBA1 clonal burden and explored its applicability in patients with VEXAS.

In a first methodological part, 6 serial DNA dilutions (10² to 10^-3 plus a set of negative control) for each of the 3 UBA1 Met41 hotspot substitutions (Leu/Val/Thr) were generated in quadruplicates (N=216 samples) and independently analyzed by ddPCR in ours and 2 external laboratories in a blinded fashion. Results were then gathered to evaluate the variability of measurements across different centers, mutations, and dilutions. For each dilution, acceptable lower and upper limits were calculated on the logarithmic scale, with VAF considered out of range if not falling within the defined limits for the respective dilutions. Intraclass correlation coefficient (ICC) was then computed as an index of interrater reliability of VAF data. F-test and confidence interval were provided on the hypothesis (r>0.5).

Overall, we observed a highly significant ICC across centers (p=7.24e-24, 95%CI=0.89-0.95). Such strong significance held true also when looking individually at the three different assays (Met41Leu, p=1.22e-19, 95% CI=0.95-0.99; Met41Thr, p=1.17e-11, 95%CI=0.87-0.96; Met41Val, p=2.85e-06, 95%CI=0.79-0.95). With regards to VAF, the assay reached a sensitivity of 10⁻² with loss of significant ICC for the 10^-3 dilution. Once assessed the reliability of our method, we asked ourselves whether longitudinal UBA1 clonal dynamics may be clinically useful to unravel interactions with specific hematological and rheumatological treatments, potentially informing therapeutic decision. To do so, we leveraged our established nationwide network and gathered N=64 samples from N=34 patients with VEXAS and the canonical Met41 hotspots at various treatment time points (N=14 onset, N=46 follow-up). All patients were initially diagnosed by means of Sanger sequencing with primers covering the entire UBA1 exon 3 from DNA extracted from either peripheral blood (PB, N=12, median VAF 46.6%, IQR=33.9-69.7) or bone marrow (BM, N=5 median VAF 92.3%, IQR=72.5-94.1). Notably, the lowest VAF in diagnostic samples was 11.9%, still amenable to be detected by means of Sanger sequencing, which in fact had a Positive Predictive Value of 100% in our cohort. However, we noticed that BM yielded higher VAF in a set of 4 patients whose paired BM/PB samples were evaluated at the same timepoint (median VAF 65.7% vs 18.5%, respectively; p=0.2). Finally, we observed highly diverse clonal dynamics according to the therapy received in samples collected longitudinally by quantifying the VAF changes across consecutive timepoints (t) ∑min(VAFt+1-VAFt ⁄ VAFt). Specifically, erythropoietin, steroids, cDMARDS (e.g.,hydroxychloroquine/methotrexate) and bDMARDs (e.g., anti-IL1/6) minimally influenced UBA1 clonality (ΔVAF= -9.4%) while ruxolitinib tended to increase it (ΔVAF= +61%). Conversely azacitdine was able to dramatically abate UBA1 clonality (median VAF 1.3%, IQR 0.3-4.8 after a median of 4 cycles, ΔVAF= -97.5%) and a complete eradication was noticed after allo-HCT, paralleling the clinical reversion of VEXAS phenotypes.

In conclusions, ddPCR for UBA1 Met41 hotspots is a reliable and reproducible method to longitudinally track clonal dynamics of patients with VEXAS. Given the different treatment interactions, this method may allow for informed therapeutic decisions and implementation of personalized strategies in a MRD-like fashion.