LGL Versus MGUS: Clonal Immune Pathogenetic Culprits in Hyporegenerative Anemias and Red Cell Aplasia

The co-occurrence of large granular lymphocytic leukemia (LGL) and plasma cell dyscrasias (PCD), chiefly MGUS, has been reported, including in our large series1. However, the epidemiologic nuances and clinical implications of this dyad warrant further study. While investigating this association, we noted a remarkable incidence of cytopenias, especially red cell aplasia (RCA). Small series have reported cases of RCA with sole LGL or PCD. Here we aimed to study cytopenias in both immune clonopathies, with a focus on the LGL/PCD/RCA triad, as a means to shed light on the proverbial deus sive natura conundrum and elucidate on pathogenetic culprits, as well as on a proper immunosuppressive therapy (IST).

Clinical review of 388 LGLs identified 97 (25%) cases with PCD; reverse screening of 2348 PCDs with marrow exam revealed 42 LGL (1.8%) cases. When opposed to sole PCD, the dyad was enriched in IgM (27 vs 15% P<.01) and small/transient M proteins (27 vs 2% P<.01). As compared to sole LGL, these cases had a higher proportion of NK-LGL (12 vs 3% P<.01). After excluding myelomas, multivariable analysis confirmed that not only LGL implied a higher risk of severe cytopenia in MGUS (OR 22 P<.01), but also MGUS in LGL (OR 2 P=.04). Evaluation of the type of cytopenia revealed that LGL increased the risk of anemia (OR 15 P<.01), neutropenia (OR 39 P<.01) and immune cytopenia (OR 14 P<.01) in MGUS, while MGUS increased only the risk of anemia (OR 2 P=.01) and immune cytopenia (OR 2 P<.01) in LGL.

Overall, we identified 40 RCA in our LGL and PCD cohorts: 35 had LGL, 27 had PCD, and 13 had both, accounting for a prevalence of RCA of 10% in LGL, 0.60% in PCD, and 13% in LGL/PCD. In the PCD setting, concurrent LGL strongly increased the risk of RCA (OR 30 P<.01). In patients with LGL, PCD also seemed to increase the risk of RCA, though with lesser magnitude (OR 2 P=.11). Median age of the LGL/PCD/RCA cohort was 74 y (57-82); F:M ratio was 1:3. The LGLs were mostly of T-cell subtype (n=12 92%); STAT3 was mutated in 7 (54%) patients. As for the PCD, the M protein was IgG in 8 (61%), IgM in 4 (31%), and IgA in 1 (8%) cases. Concurrent B-cell neoplasms were identified in 3 (23%) patients and CHIP mutations in 4 (31%) cases. Comparison of LGL/PCD/RCA vs LGL or PCD, w/o RCA, did not reveal distinct clinical or molecular patterns. We then evaluated response to IST. Frontline overall response rate (ORR) was 23%, lower than that of LGL/RCA (38%) or PCD/RCA cases (54%). For those LGL/PCD/RCA relapsed/refractory (R/R), 4 (40%) cases received bortezomib. The ORR was 75% (n=3/4). One R/R patient had a long lasting CR. This response rate was like that of our PCD/RCA cohort treated with B-cell agents (ORR 60%, CR 30%).

Further we explored mechanisms contributing to RCA in LGL/PCD. Serum IL-6 levels, measured in 33 LGL patients, peaked in LGL/PCD/RCA vs LGL/other cytopenia w/o PCD vs asymptomatic LGL w/o PCD (median: 10 vs 4 vs 2 pg/mL, P=.05). We also performed CFU assays using healthy donor bone marrow cells incubated with sera from a) LGL/RCA w/o PCD; b) LGL/other cytopenia w/o PCD; c) RCA but no LGL w/o PCD (n=18, n=3 each); d) control AB individuals (n=5). As compared with AB sera, the sera of LGLs had a negative effect on both CFU-E and GM; while the sera of RCAs selectively inhibited CFU-E. LGL/PCD/RCA sera yield the lowest normalized (% of controls) CFU-E count (44 vs 46% in LGL/RCA vs 78% in LGL/other cytopenia, P=.05). Interestingly, the serum of the only LGL/PCD/RCA with CR to bortezomib produced a real RCA in vitro, abolishing CFU-E growth. The rest of sera from RCA patients, including a case R/R to bortezomib, only produced partial inhibitions. Altogether, in the patient with CFU-E absence who showed CR to bortezomib, the M protein may have had cross reactivity with elements of the EPO axis, needed for erythroid differentiation in vitro.

To sum up, the association between LGL and PCD is more than a coincidence. Both LGL and PCD are associated with RCA. Despite LGL having a higher etiologic fraction than PCD, this may be counterbalanced at a population scale by the greater prevalence of PCD over LGL. The immune attack against erythroid precursors comes from multiple flanks and armamentaria, from TCR recognition to inhibitory cytokines. However, in the setting of RCA and T-LGL, concurrent PCD should not be obviated since it may be a wolf in sheep's clothing, sort of a monoclonal gammopathy of clinical significance. In R/R to conventional IST, antimyeloma agents may constitute promising therapeutic alternatives.

Carraway:BMS: Membership on an entity's Board of Directors or advisory committees; Servier: Membership on an entity's Board of Directors or advisory committees; Celgene: Research Funding; Novartis: Membership on an entity's Board of Directors or advisory committees; Daiichi: Membership on an entity's Board of Directors or advisory committees; Stemline: Membership on an entity's Board of Directors or advisory committees; Abbvie: Membership on an entity's Board of Directors or advisory committees; Jazz: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau.