Disseminated tumor cells in bone marrow of cancer patients : biology and clinical relevance

Molecular determinants of metastatic spread

Keratins are currently the standard markers for detection of epithelial tumor cells in mesenchymal organs such as BM, blood, or lymph nodes [ ]. Hematopoietic cells and BM stroma cells can be a source of false-positive findings but it appears that most keratin-positive cells in BM and blood samples are of epithelial origin, as indicated by the analysis of large cohorts of noncancer control patients [ ]. The most important question whether these keratin-positive cells are indeed tumor cells was answered using whole-genome amplification and subsequent DNA analysis of single DTCs [ ]. Most keratin-positive cells show genetic changes, clearly indicating that the cells are tumor cells.

However, DTCs in patients with breast cancer (BC) and other solid tumors (e.g., esophageal cancer [ ]) did not usually contain the same genetic changes as the primary tumor [ ], suggesting that DTCs may disseminate early from their primary tumor and undergo an independent genetic progression [ , ]. However, conventional CGH analysis used in these experiments is an insensitive technology with a low resolution. LOH analyses of specific genomic regions showed that genetic aberrations of CTCs in early-stage prostate cancer (PC) patients are identical to those in distinct, even small, areas of the primary tumor [ ]. A similar finding was recently observed in colorectal cancer (CRC) patients; using targeted next-generation sequencing, most CTC mutations were also revealed in small subclones of the corresponding primary tumors and metastases [ ]. These findings challenge the parallel progression theory and suggest that a metastatic subclone might already exist in the primary tumor [ ].

Recently, genome-wide copy number profiles showed that EPCAM ^positive DTCs are less aberrant than the corresponding primary tumors [ ], supporting the view that tumor cell dissemination might be an early event [ , ]. DTCs have been already found in patients with ductal carcinoma in situ (DCIS) [ ]. However, it remains questionable if tumor cells even disseminate from premalignant or benign lesions in patients, as suggested by data from some experimental mouse models [ ]. In clinical practice, patients with these lesions by definition never develop metastasis. Moreover, the hypothesis that small tumors are more actively releasing CTCs than larger tumors [ ] is contrary to the clinical observation that increasing tumor size is associated with an increased risk of metastasis and unfavorable outcome in all types of human solid tumors. Thus, cross-validation between experimental models and clinical data is crucial.

Expression profiling of DTCs can provide information on particular genes or pathways related to tumor cell dissemination to BM. Recent expression profiling of EpCAM-positive DTC pools revealed two expression subtypes: ( i ) luminal with dual epithelial–mesenchymal properties (high ESR1 and VIM/CAV1 expression), and ( ii ) basal-like with proliferative/stem cell–like phenotype (low ESR1 and high MKI67/ALDH1A1 expression). A high discordance between ESR1 (40%) and ERRB2 (43%) expression in DTC pools vs . the clinical ER and HER2 status of the corresponding primary tumors was observed, suggesting that dissemination to and survival in the BM requires a special phenotype different from those required for survival and growth at the primary tumor site.

So far, little is known about the selection of CTCs that can home to BM and become DTCs. Comparison of expression profiles of DTCs and CTCs of metastatic BC patients revealed gene expression signatures in DTCs that were unique from those of CTCs [ ]. For example, ALDH1A1 , CAV1 , and VIM were upregulated in DTC pools relative to CTCs.