Immune biomarkers of increased risk of infection in multiple myeloma (MM)

Abstract Background: Infections are one of the leading causes of morbidity and early mortality in MM. Preventing infections is paramount and immune profiles could reflect the cumulative effect of host, tumor and treatment-related immunosuppression. However, current understanding of how immune dysfunction is associated with infection risk is limited. Aim: Characterize immune profiles throughout MM progression and identify immune biomarkers of increased infection risk. Methods: The study included2,150 patients (pts). The Spanish cohort included monoclonal gammopathies of undetermined significance (MGUS, n=47) smoldering (SMM, n=374), newly-diagnosed (NDMM, n=743) and relapsed/refractory MM (RRMM, n=373). Transplant-eligible and ineligible NDMM pts were treated according to the GEM2012MENOS65 (VRd) and CLARIDEX (Rd) trials. Among RRMM pts, 149 received CD38, BCMA or GPRC5D targeting agents. Immune profiling was performed at baseline in 1,958 bone marrow (BM) and peripheral blood (PB) samples using NGF. Ten cell types were systematically quantified in BM and PB after removing tumor plasma cells (PC) from the differential (naïve and mature B cells, CD27- and CD27+ T and NK cells, basophils, eosinophils, monocytes and neutrophils). Another 5 subsets were analyzed in BM (hematopoietic progenitors, erythroblasts, mast cells, B-cell precursors and normal PC). Healthy adults of similar age (n=24) provided a reference of normal BM immune profiles. Immune biomarkers of increased infection risk identified in the Spanish cohort were validated in three external datasets: 36 pts with high-risk MGUS or low-risk SMM treated with daratumumab in the Dana-Farber D-PRISM trial, 125 NDMM frail pts enrolled in the HOVON-143 trial treated with daratumumab/ixazomib/dexamethasone, and 424 NDMM Greek pts treated with standard regimens. Results: Infection risk (per CTCAE 5.0) progressively increased in the Spanish cohort of MGUS, SMM, NDMM and RRMM (13%, 19%, 47%, 54%, p<.001). Grade 3/4 infections also increased from SMM to NDMM and RRMM (10%, 22% and 24%, p<.001). When compared to normal BM immune profiles, pts showed a higher CD27-/CD27+ T-cell ratio, eosinophils and neutrophils were reduced, while monocytes and erythroblasts were increased. Of the 15 cell types in BM, 14 had significantly different distribution across disease stages. Progression from MGUS to SMM was characterized by a reduction in B-cell precursors and normal PC, and increase in CD27- T cells. Progression to NDMM was associated with further reduction of B-cell precursors, naïve B cells and normal PC. RRMM was characterized by further reduction of memory B cells and increased CD27-/CD27+ T-cell ratio. There was an expansion of monocytes in NDMM and RRMM, mast cells were significantly increased in RRMM, while basophils and erythroblasts were reduced in NDMM. Hematopoietic progenitors progressively decreased from SMM to RRMM. Of note, these differences were observed despite similar age across disease stages, suggesting that immune profiles may better reflect immunological aging than chronological age. In the Spanish cohort, pts with any grade infection had fewer memory B cells and CD27- NK cells, as well as a higher CD27-/CD27+ T-cell ratio. These biomarkers were modeled to stratify pts with ≤1 vs ≥2 risk factors, which was associated with increased risk of infection (37% vs 60%; p<.001). Hemoglobin levels also associated with infection and non-IgA isotype was borderline significant. A logistic regression including these factors together with disease stage and CD38, BCMA or GPRC5D targeting agents unveiled that having NDMM or RRMM and ≥2 immune risk factors had independent predictive value. In the D-PRISM cohort, fewer memory B cells associated with infection risk. In the HOVON-143 trial, a higher CD27-/CD27+ T-cell ratio associated with infection risk. In the Greek cohort, fewer memory B cells and CD27- NK cells associated with infection risk. Moreover, Greek pts having ≤1 vs ≥2 of the immune risk factors identified in the Spanish cohort respectively showed 38% vs 52% infection risk. Conclusion: This is one of the largest studies analyzing immune profiles in MM. We described the effects of malignant transformation and treatment, and developed an immune score that independently predicts infection risk. Of note, the percentages of all cell types in BM and PB were significantly correlated, suggesting that these immune biomarkers could be monitored using minimally-invasive methods.