M41 and Non-M41 UBA1 Variants Differ in Their Alteration of Hematopoiesis

Background: UBA1 variants have recently been described as causative to the somatic hematoinflammatory disease VEXAS Syndrome. Previously, the features of M41 and non-M41 variants regarding their diagnoses and inflammatory transcriptome in comparison to other patients (pts) diagnosed with hematological malignancies (HM) have been described (Leukemia 37, 1080-1091, 2023). Given that UBA1 impairment also affects hematopoiesis with no evidence of malignancy, we characterized the effect of UBA1 variants on hematopoiesis including symptomatic pts without malignant diagnoses.

Aim: (1) Evaluate the hematological phenotype of UBA1 M41 and non-M41 variants detected in an unbiased cohort of symptomatic pts; (2) compare the transcriptomic profiles of M41 and non-M41 variants with healthy and pathogenic hematopoiesis.

Patients and method: Screening of 12,643 pts suspected of HM for UBA1 variants (complete coding region) by targeted sequencing. Patients were diagnosed with HM based on 2022 WHO Classification. Whole transcriptome sequencing was performed for 1297 samples (Control n = 56, CCUS n = 218, MDS n = 771, CMML n = 228, UBA1 M41 n = 16, non-M41 n = 8). UBA1 samples were selected based on availability of sample material with preference given to pts without co-mutations. K-means clustering followed by filtering with consensus score (>0.6) by ConsensusClusterPlus was used to define transcriptomic clusters. DEG was defined as |logFC| > 1.5 and adjusted p < 0.05. ToppFun was used for GO analysis with default settings.

Results: We identified 116 pts harboring UBA1 putative somatic variants (0.9%), consisting of 64 pts (0.5%) harboring M41 variants (M41 pts) and 52 pts (0.4%) harboring non-M41 UBA1 variants (non-M41 pts). As known, the majority (60/64 [94%]) of M41 pts were only suspected of HM (n = 51) or were diagnosed with MDS (n = 9) and other WHO entities such as MM were the minority. The tendency was weaker in non-M41 pts (36/52 [69 %], consisting of suspected HM n = 25, MDS n = 11; 16/52 [31%], other WHO entities). Of interest, ring sideroblasts (RS) were observed in 9/36 (25%) M41 pts and 11/28 (39%) non-M41 pts with staining available. SF3B1 was negative in 7/9 M41 pts and 8/11 non-M41 pts. Furthermore, RS ≥ 15% were observed in 4 non-M41 pts, two of which had no other co-mutations, suggesting the involvement of non-M41 variants in RS generation.

To understand the effect of UBA1 in hematopoiesis, we compared the expression of marker genes of hematopoiesis of M41 and non-M41 pts with healthy individuals, pts diagnosed with CCUS, MDS, or CMML. The reference groups were constructed by k-means clustering based on 212 genes standardly recognized as relevant in hematopoiesis, and pts with the strongest differentiation bias are shown (Figure). The known myeloid bias and B-cell decrease of M41 pts were observable, and half of the M41 pts had at least a weak membership to clusters with some myeloid bias (Cluster 2 and 4). The observed myeloid bias could either be granulocytic or monocytic and the tendency was not homogenous among M41 pts. In contrast, half (n = 4) of the non-M41 pts belonged to the erythroid bias Cluster 5, which is enriched in pts carrying SF3B1 variants, and 3 belonged to Cluster 5 without SF3B1 variants. Neither M41 nor non-M41 pts clustered with transcriptomic groups of uni-lineage differentiation bias (Cluster 1, 3 and 6), and only one non-M41 pt clustered with the healthy controls, suggesting low risk but pathogenic hematopoiesis.

To further understand the erythroid bias in non-M41 variants, we performed DEG and GO analysis. As erythroid bias comparators, we chose male pts in Cluster 5 diagnosed with MDS-SF3B1 (n = 14) and Cluster 6 diagnosed with MDS-biTP53 (n = 8), and male controls (n = 30). The pt with a SF3B1 variant was excluded from the non-M41 group. Compared to controls, both the non-M41 group and Cluster 5-SF3B1 had upregulated heme synthesis, sharing upregulation of SF3B1-targets ABCB6 and PPOX. Cluster6-biTP53 was more characterized by the downregulation of myeloid genes, which was not shared with non-M41 pts. Prognostic marker WT1 was upregulated in non-M41 group compared to Cluster 5-SF3B1 but not to controls or Cluster6-biTP53.

Conclusions: UBA1 M41 variants confer myeloid bias but do not lead to the uni-lineage differentiation bias seen in pts with high risk mutations. Isolated UBA1 non-M41 pts can have RS ≥ 15%, and manifest erythroid bias similar to MDS-SF3B1 but higher expression of prognostic marker WT1.