Immune Biomarkers of Survival and Severe Infection in Newly Diagnosed Multiple Myeloma (NDMM) Patients (pts) Treated with the Backbone Regimen Lenalidomide and Dexamethasone (Rd)

Introduction: Rd is a backbone of treatment in MM. The strong immunomodulatory effects of this regimen would suggest that immune profiling might complement conventional prognostic factors. However, immune signatures predictive of survival were not well defined in the past, and nowadays it is challenging to detach the effect of Rd from other backbone drugs in triplets or quadruplets, as well from high dose therapy in transplant eligible patients.

Aim: Identify immune biomarkers in elderly NDMM pts treated with Rd.

Methods: The GEM-CLARIDEX trial randomized transplant ineligible NDMM pts to Rd +/- clarithromycin until disease progression. Because clarithromycin had no effect in survival, the trial design yielded a unique opportunity to identify immune biomarkers in elderly MM pts treated with Rd. Immune profiling was performed in bone marrow (BM) at baseline in 186 pts, using multidimensional and computational flow cytometry together with lasso penalized Cox regression and 10-fold cross validation, for a holistic and unbiased definition of prognostic immune signatures. 38 cell populations were identified, including seven myeloid/erythroid and 29 lymphoid subsets plus normal and clonal plasma cells (PC). Single cell RNA sequencing (scRNAseq, 10x Genomics) was performed on BM immune cells sorted from healthy adults, MGUS, smoldering and active MM pts (n = 17).

Results: An immune signature associated with inferior survival was defined based on increased frequency of cytotoxic CD314⁺ NK and CD4⁺ central memory (CM) CD127⁺ PD1⁺ T cells, and decreased percentages of CD4⁺ CM CD127⁺ PD1^- and CD8⁺ CD127^low PD1⁺ terminally differentiated T cells, as well as of B cell precursors. Median progression-free survival (PFS) of NDMM pts with high risk immune score (106/186 [57%]) was 18 months vs not reached in those with low risk (80/186 [43%]) (p=.0008, Fig 1A). The respective median overall survival (OS) rates at three years were 50% and 80% (p = .0021, Fig 1B). In multivariate analyses including age, ISS, LDH and cytogenetic risk, immune profiling showed independent prognostic value for PFS (HR: 2.6, p = .001) and OS (HR: 2.9, p = .003).

Increased frequency of cytotoxic CD314⁺ NK cells was the immune biomarker with the highest negative impact in the survival predictive model. In an attempt to better understand this finding, we analyzed scRNAseq data of 61,637 NK cells from normal and tumor BM. Up to 47 NK-cell clusters were identified, and the one showing the highest expression of CD314 also displayed upregulation of granzymes K, H and B, perforin, CD16, CD38, CD69 and IFNγ, and therefore appeared to be associated with cytolytic activity and inflammation. Interestingly, there was a progressive expansion of this cluster within the NK cell compartment of normal vs pre-malignant vs neoplastic BM samples. Similarly, in genetically engineered BIcγ1 mice that develop active MM preceded by precursor states, there was a progressive expansion of CD314+ NK cells in normal vs benign vs malignant BM (p = .001).

Being infection an important cause of morbidity and mortality in elderly MM pts as those included in this study, we next investigated if immune profiling at diagnosis could help predicting severe infection (≥ grade 3 CTCAE v4.0, n=43/186). The combination of five clinical features – age ≥ 75y, low hemoglobin and albumin, high β2m and LDH – plus seven immune biomarkers – increased percentages of eosinophils, together with decreased frequencies of CD4⁺ CM CD127⁺ PD1^- _, CD8⁺ naïve and CD127^low PD1⁺ and PD1^- terminally differentiated T cells, B cell precursors and normal PC – resulted in the most effective model to predict severe infection. B cell precursors and eosinophils, which are easily identified using flow cytometry at diagnosis, showed the strongest association. The area under the curve (AUC) of the model was 0.90 with a 4-fold cross validation showing a mean AUC = 0.86 ± 0.06.

Conclusions: This study shows the clinical value of immune profiling at diagnosis and defined immune cell types, including a previously unidentified subset of CD314⁺ cytotoxic NK cells, which were predictive of survival in transplant ineligible NDMM pts treated with the backbone regimen Rd. Furthermore, this study uncovered immune biomarkers complementary to clinical features for predicting severe infection in elderly MM pts, which could be used to intensify prophylactic measures in specific subgroups that are more susceptible to severe infection.