Abstract
Introduction Bone marrow biopsies (BMBx) are an integral part of the diagnosis and management of plasma cell disorders (PCD) because they provide access to the tumor cells themselves but also to the immune microenvironment relevant to these disorders. However, due to the invasiveness, discomfort, and expense, BMBx are only utilized at a very limited number of timepoints during diagnosis and treatment. With the evolving role of liquid biopsy techniques in the management of hematologic malignancies, a deeper understanding of how closely the peripheral blood (PB) immune milieu reflects that of the BM, has the potential to add another dimension to the pursuit of less invasive, less costly methods for diagnosis and monitoring disease response. Furthermore, immune recovery is a new area of interest in patients having received effective therapies, but the timepoint for the assessment is unclear. PB tests allow for more frequent and less invasive testing than BMBx. The current study provides insight into which lymphocyte populations and functional subsets are most closely correlated in paired PB and BM.
Methods The patients included in this analysis (n=65) were being worked up for PCDs and were ultimately diagnosed with multiple myeloma (n=36), smoldering multiple myeloma (n=5), monoclonal gammopathy of undetermined significance (n=11), Waldenström macroglobulinemia (n=4), light chain amyloidosis (n=3), or another hematologic condition (n=6). Sampling of paired PB and BM were done within 15 days of one another and were done on treatment-naïve patients. A comprehensive flow cytometry panel was utilized to assess the PB and BM frequency of lymphocyte populations, including CD4+, CD8+, and gamma-delta (γδ) T cells, B cells, natural killer (NK) cells, and NKT cells. The current analysis includes 47 markers of interest representing absolute cell counts (per μL) of these populations and major functional subsets, such as naïve, memory, exhausted, and terminally differentiated, as well as key ratios of interest. Correlations between populations in PB and BM are estimated using Spearman's rank correlation coefficients (ρ), with p-values ≤.05 considered significant. The false discovery rate (FDR) is controlled using the Benjamini-Hochberg procedure.
Results Patients included in the study have a mean age at sampling of 67.0 years. More than two-thirds (69%) of the patients are male.
As expected, the median absolute counts of most lymphocyte populations are greater in the BM than PB, with the exception of CD4+ T cells. Overall, there is strong correlation between the paired specimens, with the median correlation of all 47 markers being ρ=.80. All p-values are <.0001, and all immune markers remain significantly correlated after correction for FDR.
There are 9 markers with very strong correlation (ρ>.90), 31 with strong correlation (ρ>.70), and 7 with moderate correlation (ρ>.50). The major lineages had the following correlations between PB and BM: CD4+ T cells (ρ=.81), CD8+ T cells (ρ=.84), γδ T cells (ρ=.85), B cells (ρ=.86), NK cells (ρ=.65), and NKT cells (ρ=.92). The functional subsets that are very strongly (ρ>.90) correlated between PB and BM are CD4+ naïve T cells (ρ=.92), naïve B cells (ρ=.91), non-class switched memory B cells (ρ=.91), NKT cells (ρ=.92), terminally differentiated NKT (ρ=.95), and the following ratios: CD4/CD8 (ρ=.94), CD4+ naïve/memory (ρ=.94), CD4+ exhausted/non-exhausted T cells (ρ=.92), and CD8+ exhausted/non-exhausted T cells (ρ=.93). The populations with the lowest correlations between PB and BM include CD4+ effector memory (TEM) (ρ=.69), CD4+ terminally differentiated effector memory (ρ=.68), CD4+ regulatory T cells (ρ=.65), CD8+ TEM (ρ=.66), the ratio of CD8+ TEM to central memory (TCM) (ρ=.59), class-switched memory B cells (ρ=.68), and NK cells (ρ=.65). Symptomatic and asymptomatic patients had similarly strong correlations between PB and BM, with the exception of CD4+ TIGIT and PD1 exhausted populations and CD8+ TEM being more strongly correlated in asymptomatic, and CD8+ naïve and B cells being more strongly correlated in symptomatic patients.
Conclusion The current study presents evidence that the PB immune profile is strongly correlated with that of BM. Additional work is needed to study rarer and more specifically defined populations, to examine how these relationships hold up in different PCD diagnoses, and to examine whether treatment impacts these PB-BM correlations.
