IDH-mutant myeloid neoplasms are associated with seronegative rheumatoid arthritis and innate immune activation

TO THE EDITOR:

Autoimmune manifestations including autoimmune rheumatic disease (AIRD) are reported in 8% to 30% of patients with myeloid neoplasm (MN), including myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML), acute myeloid leukemia (AML), and myeloproliferative neoplasms (MPNs).^1-4 Furthermore, the risk of developing MN is elevated in autoimmune diseases.^5,6 However, the mechanisms underpinning the association between autoimmune disease and MN remain unclear.

We assessed the burden of AIRD in patients with MN (n = 1702) including MDS (n = 861), AML (n = 640), MDS/MPN (n = 112), and MPN (n = 89). AIRD was identified using the International Classification of Diseases, which was then verified by experienced rheumatologists (supplemental Figure 1A-B and supplemental Table 1, available on the Blood website). The study was approved by the respective ethics committees and performed in accordance with the Declaration of Helsinki. Methodological details are provided in the supplemental materials.

The median age of the cohort was 68 years (interquartile range, 59-75), 1023 (60%) of the subjects were male (supplemental Table 2), and 7.7% of patients with MN had AIRD (MN-AIRD). Enrichment of AIRD was observed in MDS, MDS/MPN, and MPN compared with AML (9.5%, 9.8%, 10.1%, and 4.7%, respectively; P = .003) (Figure 1A). MN cohort displayed male predominance (60.1% vs 39.9%), but a higher proportion of females had concomitant MN-AIRD (55.7% vs 44.3%; P < .01). This was confirmed when analyzed by sex (11% female vs 6% male, P < .0001), especially in MDS-AIRD (supplemental Figure 1C-E). Inflammatory arthritis (n = 64, 48.9%), inflammatory connective tissue diseases (n = 23, 17.2%), polymyalgia rheumatica (n = 18, 13.7%), and vasculitis (n = 16, 12.2%) were the most prevalent AIRD (Figure 1B). Notably, 85.9% of inflammatory arthritis diseases were rheumatoid arthritis (RA) (n = 55) followed by peripheral (n = 8, 12.5%) and axial (n = 1, 1.6%) spondyloarthropathies.

Enrichment for somatic mutations in the epigenetic modifiers DNMT3A (odds ratio 1.93, P = .02) and IDH1/2 (odds ratio 1.9, P = .02) was observed in MN-AIRD (Figure 1C). To investigate this observed link, we analyzed an expanded cohort of 1356 patients with MN screened for IDH1/2 mutations. Enrichment of IDH1/2 mutations was identified in all MN-AIRD cases compared with non-AIRD cases (18.8% vs 11.3%; P = .04) driven primarily by the enrichment noted in MDS-AIRD (20.7% vs 6.9%; P = .001), suggesting a link between these somatic mutations and autoimmune features (Figure 1D). A similar trend was observed in AML (22.7% vs 18.7%; P = .58) but did not reach statistical significance in part due to the higher prevalence of seropositive RA in AML compared with MDS (78% vs 29%; P = .03), and the lack of association of seropositive RA with IDH mutation. Unexpectedly, males with MN-AIRD displayed IDH1/2 mutations enrichment when compared with those without AIRD (23% vs 11%; P = .01), which was not apparent in females (14% vs 12%; P =.67) (Figure 1E).

We next evaluated AIRD subtypes in patients with IDH1/2 mutations. IDH1/2 mutations were increased in MN with RA vs other AIRD (30.6% vs 11.7%; P = .03) (Figure 1F-G) with striking enrichment in seronegative vs seropositive RA (44.4% vs 0%; P = .009) (Figure 1H). Although IDH1/2 mutations occurred with similar frequency in AIRD, we found enrichment of IDH1 mutations in seronegative RA (75% IDH1 vs 25% IDH2), and IDH2 mutations were prevalent in other AIRD (100% IDH2 vs 0% IDH1; P = .009) (Figure 1I). The higher frequency of AIRD was verified in an independent validation cohort of patients with CMML (n = 21) who were selected based on IDH mutation status, irrespective of AIRD status. Of the patients with IDH1/2-mutated CMML, 62% (n = 13) had AIRD (n = 7) or other autoinflammatory manifestations (n = 6) (Figure 1J). Our findings indicate enrichment of IDH1 and IDH2 mutations in AIRD, particularly seronegative RA.

If IDH1/2 mutations play a pathogenic role in AIRD, it is plausible that mutant clones precede AIRD diagnosis. We therefore analyzed the temporal events surrounding mutation and disease onset. In 84% (n = 65) and 5% (n = 4) of cases with available information (n = 78), MN was diagnosed >12 and 1 to 12 months after the AIRD diagnosis, respectively, and in 6% (n = 5) and 5% (n = 4) of cases, MN was diagnosed within 1 month and >1 month before the AIRD diagnosis, respectively. The interval between diagnosis of RA and MN (latency) tended to be shorter in seronegative vs seropositive RA (34.6 vs 135.4 months; P = .06) (supplemental Figure 1F). Importantly, the interval from AIRD to MN was markedly shorter in IDH1/2-mutant cases compared with IDH wild-type cases (44.4 vs 106.3 months; P = .02) (Figure 2A) and even shorter in patients with IDH1 vs IDH2 (12.9 vs 85.6 months; P = .04) (Figure 2B). This short latency between AIRD and MN could be explained by the clonal expansion rate of 20% and 10% per year of IDH1- and IDH2-mutated clones, respectively, which is faster than the yearly expansion rate of 5% for DNMT3A and TP53 clones.⁷ Furthermore, extrapolation of the IDH1/2-mutant clonal burden prior to MN diagnosis indicates that the majority of clones are expected to be detectable at the time of AIRD diagnosis (Figure 2C).

AIRD are thought to be driven through adaptive immunity. However, autoinflammatory disorders such as VEXAS (vacuoles, E1 enzyme, X linked, autoinflammatory, somatic) syndrome are associated with innate immune activation and MDS.⁸ We therefore analyzed T-cell subsets and monocyte and macrophage function in AIRD vs non-AIRD MN. We found a higher ratio of proinflammatory CD14⁺CD16⁻ classical monocytes to nonclassical anti-inflammatory CD14⁻CD16⁺ monocytes in MN-AIRD compared with age-matched controls and cases of MN without AIRD (Figure 2D; supplemental Figure 2). In contrast, no difference in T-cell subsets (including naive, central memory, effector memory, and terminally differentiated CD4⁺ and CD8⁺ T cells) was observed (supplemental Figure 3). Furthermore, we observed increased levels of proinflammatory cytokines, including granulocyte-macrophage colony-stimulating factor, interleukin-12 (IL-12), fractalkine, IL-15, and IL-1β, in the bone marrow plasma of patients with MN-AIRD vs those without AIRD (Figure 2E). Consistent with these cytokines being predominantly derived from innate immune monocyte/macrophages rather than T cells, we observed increased monocyte-derived phagocytic activity (assessed by flow cytometry) in MN (Figure 2F).

We then isolated CD14⁺ monocytes, CD3⁺ T cells, and CD19⁺ B cells from 2 patients with AIRD with known IDH1/2 mutations (Figure 2G). Interestingly, high VAF clones were detected in monocytes (40%-50%) but not in T cells (<1%) or B cells (<1%) from both patients (Figure 2G; supplemental Figure 4), consistent with intracellular (R)-2-hydroxyglutarate (2-HG) production in monocyte/macrophage cells rather than T cells, leading us to assess 2-HG levels in patients with AIRD. Strikingly, all 3 cases with high 2-HG (bone marrow metabolite peak abundance above median of >9.3 × 10⁵) had AIRD compared with only 1 of the 5 cases with low 2-HG (<9.3 × 10⁵ median abundance, P = .02) (Figure 2H). Finally, we noted significantly higher macrophage phagocytic activity in subjects with IDH1/2-mutant MN compared with controls (Figure 2I).

IDH mutations are neomorphic and lead to the production of an alternative metabolite in the citric acid cycle, 2-HG, which inhibits TET2 deoxygenase with resultant genomewide epigenetic modifications. This study demonstrates the enrichment of IDH^mut in MN-AIRD, in line with previous findings.⁴ Importantly, our study demonstrates a strong association between IDH-mutated MN and a specific type of AIRD-seronegative RA, as well as a temporal relationship between IDH-mutant clones and AIRD. The observed elevated 2-HG in seronegative RA and IDH1/2 mutations in myeloid rather than T cells indicates a causative association between IDH1/2 mutation and AIRD. Emerging literature from in vitro⁹ and in vivo studies reports proinflammatory effects of IDH mutations on monocytes⁹ and macrophages,¹⁰ in addition to an elevated IL-18 level in individuals with IDH1 clonal hematopoiesis.¹¹ Collectively our findings suggest a causative relationship between IDH1/2 mutations, MN, and seronegative AIRD, meriting further exploration of mutant IDH inhibitors such as ivosidenib,¹² enasidenib,¹³ or complex I inhibitor¹⁴ as therapeutic options for seronegative RA.