Session: 652. Multiple Myeloma and Plasma Cell Dyscrasias: Clinical and Epidemiological: Precursor States: MGUS and Smoldering Myeloma
Hematology Disease Topics & Pathways:
Research, Translational Research, Plasma Cell Disorders, bioinformatics, Diseases, immunology, computational biology, Lymphoid Malignancies, Biological Processes, Technology and Procedures, Human, Study Population, omics technologies
Patients with Smoldering Multiple Myeloma (SMM) have altered immune cell composition in the bone marrow (BM) and show inferior responses to SARS-CoV-2 vaccination, which indicates defective immune cell function. Data from two randomized Phase III trials suggest that therapeutic intervention prior to progression can prolong progression-free survival (PFS) in patients with high-risk SMM (Mateos et al., 2013; Lonial et al., 2020). However, little is known about the role of BM-based immune profiling in risk stratification and patient selection or the utility of peripheral blood (PB) for the detection of immune alterations related to disease.
Here, we performed single-cell RNA- and T-cell receptor (TCR) sequencing on 149 samples, drawn from 34 patients with high-risk SMM who were enrolled on a Phase II clinical trial of Elotuzumab, Lenalidomide, and Dexamethasone (E-PRISM), and 21 healthy donors. Specifically, we profiled 60 CD138- BM mononuclear cell samples drawn at baseline (n=28), cycle 9, day 1 (C9D1, n=16) and end of treatment (EOT, n=16), 68 PB mononuclear cell (PBMC) samples drawn at baseline (n=33), C9D1 (n=14), and EOT (n=21), and 11 BM and 10 PBMC samples from age-matched healthy donors. For 24 patients and 40 samples, data was generated from matched BM and PB immune cells to enable head-to-head comparisons. We also integrated non-trial patients with our E-PRISM cohort, including MGUS (n=6), low-risk SMM (LRSMM, n=3), high-risk SMM (HRSMM, n=12), newly diagnosed MM (NDMM, n=9) and healthy donors (n=11), reaching a total of 190 samples and 231,608 cells. Libraries were prepared using Chromium Single-Cell 5’ and 3’ kits by 10X Genomics and sequenced on NovaSeq at the Genomics Platform of the Broad Institute of MIT and Harvard (Cambridge, MA).
First, we hypothesized that the similarity of a patient’s BM immune cell composition to that of healthy donors may have prognostic value. We trained a Naïve Bayes classifier on BM samples (n=41) from patients and healthy donors to identify the presence of SMM, which achieved 94% accuracy in the testing set (n=16). We computed normalization scores for each sample by summing the product of each cell type’s proportion and its corresponding signed importance to the classification, and classified patients based on the median as least (reactive) or most normal-like (non-reactive) at baseline. Patients classified as reactive (n=12) had significantly prolonged PFS in response to therapy (p=0.011). Furthermore, reactive patients showed significantly lower abundance of pro-inflammatory monocytes and pDCs, while their granzyme K (GZMK)-expressing CD8+ T-cells showed lower levels of exhaustion markers (including PDCD1, TNFRSF9, and TOX) and higher levels of genes associated with long-lived memory T-cells (CD27, IL7R), and functionality (IFNG, TNF). On average, patients showed significantly higher normalization scores at EOT (p=0.018). Based on the change in normalization scores from baseline to EOT, we defined a state of post-therapy immune normalization (PIN) and classified four patients who showed no or minimal increase in their normalization score at EOT as PIN-negative. Patients who achieved PIN at EOT (n=8) had significantly longer PFS (p=0.032).
Next, we tested whether PB can be used reliably for immune profiling of patients with HRSMM, as BM biopsies are invasive and carry risk, which prohibits regular sampling. Indeed, by comparing PB samples between patients and healthy donors, we recovered compositional changes discovered in our BM-based analysis and showed that the TCR repertoire of patients with SMM is significantly less diverse than that of HD in both the BM and PB. Strikingly, our BM-trained classifier was able to correctly infer the presence or absence of SMM in nearly all PB samples from patients and healthy donors with an accuracy of 97%.
Collectively, our results indicate that the similarity of a patient’s BM immune cell composition to that of healthy donors can be harnessed for prognostication both at diagnosis and post-therapy, which could have implications in risk stratification at diagnosis, trial design and monitoring of response to therapy. Moreover, PB-based immune profiling may have diagnostic and prognostic utility, potentially enabling regular assessments of immune dysregulation in patients.
Disclosures: Haradhvala: MorphoSys: Consultancy. Aguet: Illumina: Current Employment. Zavidij: Constellation Pharmaceuticals, Inc., a MorphoSys Company: Ended employment in the past 24 months. Getz: IBM: Research Funding; SignatureAnalyzer-GPU: Patents & Royalties; Scorpion Therapeutics: Consultancy, Current equity holder in publicly-traded company, Other: Founder; Pharmacyclics: Research Funding; MSMutSig: Patents & Royalties; MSMuTect: Patents & Royalties; MSIDetect: Patents & Royalties; POLYSOLVER: Patents & Royalties. Ghobrial: Sognef: Honoraria; Bristol Myers Squibb: Honoraria; GSK: Honoraria; Huron Consulting: Honoraria; Veeva Systems: Honoraria; Vor Biopharma: Honoraria; Aptitude Health: Honoraria; Sanofi: Honoraria; Menarini Silicon Biosystems: Honoraria; Window Therapeutics: Other: Advisory board participation; The Binding Site: Honoraria; Takeda: Honoraria; Pfizer: Honoraria; Oncopeptides: Honoraria; Janssen: Honoraria; Amgen: Honoraria; Adaptive: Honoraria; AbbVie: Honoraria; Novartis: Research Funding; Celgene: Research Funding.
See more of: Oral and Poster Abstracts