Biology of circulating tumor cells (CTCs) in multiple myeloma (MM): Comparative transcriptomic profiling reveals egression-associated changes related to CTC burden

Abstract

Abstract Background: CTCs have emerged as a key prognostic factor in newly diagnosed (ND) MM. However, mechanisms of plasma cell (PC) egression from the bone marrow (BM) to peripheral blood (PB) remain poorly understood. To date, differential expression comparisons between BM plasma cells (BMPCs) and CTCs have been hampered by technical limitations driven by low CTC levels precluding successful isolation of sufficient number of PCs. In this work, we address this gap in understanding of CTC biology. Aim To characterize transcriptomic differences between CTCs and paired BMPCs in NDMM using a wide range of CTC levels and to identify mechanisms underlying PC egression from BM and their survival in the PB. Methods PB samples from 29 NDMM/primary plasma cell leukemia (pPCL) patients were analyzed by next-generation flow cytometry (EuroFlow protocol) and stratified into logarithmic groups: 8 patients (10–100%), 6 (1–10%), 8 (0.1–1%), 4 (0.01–0.1%), and 3 (0.001–0.01%). Paired CTCs and BMPCs were either sorted by FACS (N=21) or by combining MACS and FACS (N=8) to obtain sufficient number of cells in low-CTC samples. Differential gene expression was analyzed using DESeq2 (paired design) with apeglm shrinkage, and pathway enrichment with clusterProfiler. Associations of gene expression with progression-free (PFS) and overall survival (OS), were assessed by log-rank tests in 635 CoMMpass NDMM patients with transcriptomic and genomic data. To derive the CTC-like score, genes upregulated in CTCs were weighted by log2FC and –log₁₀(padj). For each CoMMpass sample, the score was computed as the weighted sum of expression values. All reported p-values were adjusted using Benjamini-Hochberg method. Results We identified 1,142 differentially expressed genes, including 66 upregulated and 28 downregulated (absolute log2FC>1) in CTCs. Genes and transcriptional signatures related to proliferation (e.g., PR score; Zhan et al., 2006) were significantly lower in CTCs than BMPCs (p<0.001), consistent with reduced proliferative capacity in absence of BM niche. Notably, key BM homing molecules CXCR4 and CD138 had lower expression in CTCs. Conversely, transcriptomic classifier for pPCL (PCL-like score, Bruinink et al., 2022) was higher in CTCs than in BMPCs (p<0.001). Upregulated genes such as FLNA, TAGLN2, CD44, or EMP3 suggest altered cytoskeletal dynamics and microenvironment interactions, implying greater mechanical resilience and BM egression capacity. Notably, all 3 genes of the ANXA2-S100A10-AHNAK complex were upregulated, linked to extracellular matrix (ECM) remodeling and cytoskeletal dynamics promoting cell spreading. Importantly, differences between CTCs and paired BMPCs were more pronounced in cases with lower CTC burden. Gene expression variability in CTC samples was low across logarithmic groups, unlike BMPCs where variability was higher, especially between high- and low-CTC cases. This trend was particularly strong in 24/66 (36.4%, p<0.05) CTC-upregulated genes. The majority of these genes were associated with worse prognosis in CoMMpass, suggesting a role in disease aggressiveness. To investigate genomic alterations linked to CTC burden, we computed CTC-like transcriptional signature score for CoMMpass samples. Higher scores were associated with cytogenetic features, including 1q gain, t(14;16), and 13q deletion, and inversely with hyperdiploidy, alterations previously linked to elevated CTC numbers in CoMMpass and other datasets. Conclusion The combination of MACS and FACS techniques enabled us to isolate sufficient numbers of CTCs for transcriptomic profiling. In the majority of NDMM cases, CTCs exhibited reduced PR and increased PCL-like profile compared to their BM counterparts. While high expression of proliferation genes is typically linked to poor prognosis, their lower levels in CTCs might reflect the shift into a less supportive environment, rather than reduced aggressiveness. Transcriptional differences between BMPCs and CTCs increased with decreasing CTC burden, while BMPCs resembled CTCs more as CTC burden rises. Upregulation of ANXA2, S100A10, AHNAK, CD44, FLNA, and TAGLN2, alongside downregulation of CXCR4 and CD138, suggest that the underlying mechanisms of PC egression from BM to PB include altered cytoskeletal dynamics, ECM remodeling, and reduced BM homing. The derived CTC transcriptional score was associated with cytogenetic features, including t(14;16), 1q gain, and 13q deletion using CoMMpass database of 635 patients.