Are disseminated tumor cells in bone marrow and tumor-stroma ratio clinically applicable for patients undergoing surgical resection of primary colorectal cancer? The Leiden MRD study

In nearly one-fifth of the patients with CRC, including early stage cases, we found DTCs in the BM. These DTCs, detected by CK-ICC, were not found to impose a significant impact on the prognosis of patients that underwent resection of the primary tumor. The idea that a high amount of stroma in the primary tumor may be related to the presence of DTCs was not confirmed by our study. We did find, however, that a high stroma-percentage was associated with a worse overall survival in the CRC patients included in this study.

Results obtained by others aimed at assessing the prognostic impact of DTCs in primary CRC cases have so far been conflicting. The amount of positive DTCs in BM were e.g. found to range from 24 to 64 % [6, 17‐22]. These differences in (mostly) cytokeratin stained cells may reflect the use of different techniques. Flatmark et al. [23] found that the use of two different techniques (immunomagnetic selection with an anti-EpCAM antibody and CK-ICC) resulted in a minimal detection overlap although, surprisingly, the results of both methods were found to be associated with disease outcome in distinct CRC prognostic subgroups. The main conclusion from their study was that the presence of DTCs in BM may serve as a prognostic biomarker for curatively resected CRC patients.

In some studies, the presence of DTCs was found to be associated with a shorter disease-free survival [18, 24], whereas in other studies no prognostic relevance of DTCs in the BM was noted [6, 19, 21]. Steinert et al. [6] detected the same amount of CK-ICC positive cells throughout all CRC TNM stages, and Lindemann et al. [18] failed to find a significant difference between lymph node-positive and -negative patients, although they did find that the detection of tumor cells in the BM may serve as an independent determinant of relapse. In both studies immunocytochemistry was used for BM analysis. Leinung et al. [17] found an increased percentage of CK-ICC positive cells in the BM in rectal cancer patients compared to colon cancer patients. In 2010, Rahbari et al. published a meta-analysis (in which the above described studies were also included) to assess whether hematogeneous DTCs represent a prognostic factor in patients with CRC. They found a strong prognostic effect of circulating tumor cells in the peripheral blood but, in accordance with our findings, this effect was not found for DCTs in the BM [25].

A pooled analysis of micro-metastases in the BM of breast cancer patients [7] revealed that the presence of DTCs in the BM was predictive for the development of distant metastases after long term follow-up. Tumor cell persistence in BM was also found to be an independent prognostic factor for subsequent breast cancer survival [26]. In contrast to breast cancer, in which bone is one of the most preferential metastatic target sites affecting more than half of the patients during the course of their disease, the incidence of bone metastases in CRC is rare [27, 28]. Three of our 125 patients developed skeletal metastases during follow-up, two of which were BM-positive. In 18 % of the patients DTCs were detected in the BM, but this does not seem to reflect the presence of skeletal metastases. In our control group of 20 breast cancer patients, we found in 45 % of the cases DTCs in the BM. This result is in accordance with other studies reporting DTC occurrences in the BM of 13 to 50 % in early stage breast cancer patients, and increases in patients with metastatic breast cancer of up to 70 % [29]. In 2 of 29 non-carcinoma control cases we found CK-ICC positive DTCs in the BM. This finding is not novel. Depending on the antibody used, CK-positive cells in BM have been found in up to 5.5 % of individuals without a known malignancy [29]. Therefore, it may be questioned whether all disseminated epithelial cells in the BM are truly malignant cells. Although earlier characterization of these cells appeared to be difficult, a malignant phenotype has been suggested [20, 30]. So, the option must be considered that these cells are indeed tumor cells, but that these cells may not all have the capacity to further metastasize. On the other hand, it has recently been found in an experimental model of colon cancer induced by male cancer cells injected into female nude mice that disseminated ‘tumor’ cells (i.e., epithelial-like cells) may be composed of two different populations, one originating from the cancer (cytokeratin-positive, Y chromosome-positive) and one originating from the resident BM cells [31].

The clinical relevance of the molecular detection of DTCs appears to be restricted by both tumor cell heterogeneities [32] and technical limitations. Genomic analyses at a single cell level have shown that DTCs detected with anti-cytokeratin antibodies frequently exhibit heterogeneous tumor-specific aberrations, particularly in patients without overt metastases [33]. Many tumors are known to undergo an extended period of ‘dormancy’, but little is known about the mechanisms underlying the ‘awakening’ of the dormant tumor cells, which may ultimately lead to the formation of metastases. The steady state of dormancy may be disturbed by changes in both DTCs and their surrounding microenvironment [33]. Tumor cells in the BM of CRC patients may be in the “wrong” environment and, as such, be kept in the dormant state [21]. Future efforts towards comprehensive genomic analysis of DTCs may provide a deeper understanding of a clinically relevant biology of DTCs [34].

For many years, tumor stromal formation or desmoplasia was considered to be a passive bystander of tumorigenesis and tumor progression. However, during the last 10 years, attention has shifted towards the tumor microenvironment and it is now well established that tumor stroma plays an important role in cancer initiation and progression. The tumor stroma interacts with nonmalignant cells as well as with malignant cells during different stages of tumorigenesis, ranging from tumor onset to invasion and metastasis [35, 36]. TSR is a relatively simple to apply and cheap cell-based parameter, which is a strong quality. Multiple other studies have confirmed significant prognostic implications of the TSR not only in colon cancer, but also in other solid tumors [37‐42]. Further research using a larger prospective cohort should bring this parameter closer to implementation in the TNM classification system.