Background
Clinical use of some newer or investigational drug therapies in cancer requires that primary tumors be assayed for specific mutations associated with response or lack of response [
1‐
5]. However, not only are primary tumors known to be mutationally heterogeneous [
6‐
8], new mutations may become apparent in recurrent tumors, emerging during disease progression [
9‐
11]. Yet sequential biopsy and evaluation of molecular biomarkers and mutations in metastases is not routinely done, even during clinical trials [
11], largely due to the multiplicity and internal location of many metastases (such as liver, lung, and/or brain metastases in breast cancer), and potential morbidity associated with sequential biopsy. An appealing alternative is a “liquid biopsy” with CTC capture and characterization [
12,
13]. As they are easily accessible by simple blood draw, CTCs can be sequentially sampled at multiple time points during the course of disease for biomarker or genotype determination. Moreover, it is hoped, but not ascertained, that CTCs represent mixtures of tumor cells that reflect the full spectrum of molecular phenotypes and genotypes present in multiple metastases.
Following a slightly different tack, some groups are also investigating biomarker and genetic characteristics of DTCs from bone marrow [
14‐
16], which are postulated to serve as a reservoir for active and dormant tumor cells [
12,
17]. However, genetic analyses of mutations in CTCs and DTCs are still in an early discovery stage, having been done on few patients, so clinical significance and utility is postulated but remains unproven. Moreover, it is not known whether CTC analysis can replace DTC analysis for there is, as yet, incomplete understanding of the relationship between these two populations [
18].
In the present study, we used a previously described magnetic separation technology that isolates live single cells [
19‐
21] for mutation analysis of single cells from different compartments in metastatic breast cancer patients and also demonstrate growth in culture of patient DTCs from bone marrow. For single CTC/DTC/tumor mutation analysis, we have chosen to interrogate exons 9 and 20 of the
PIK3CA gene, one of the most frequently mutated genes in breast cancer [
22‐
25]. We demonstrate that this mutation can be detected in single tumor cells isolated from breast cancer patient primary tumor, blood, bone marrow, and metastases, and track mutational status of CTCs over time in a metastatic breast cancer case example and in cultured DTCs from this patient. While we have previously shown that individual CTCs in breast cancer, even from the same blood draw, are transcriptionally heterogeneous [
21], here we investigate mutational heterogeneity and concordance among CTCs, DTCs, and single tumor cells from primary tumors and metastases. In particular, for CTCs to be ultimately used to guide drug selection, we hypothesized that CTCs should indeed contain the mutational changes found in metastases. However, our results were surprising and we present here a case that provides a cautionary note that CTCs from any one blood draw alone may not always represent the mutational status of tumor cells in bone marrow or distant metastases.
Discussion
Investigations are underway exploring the clinical utility of CTCs and DTCs in monitoring cancer patients undergoing systemic drug therapy [
35]. However, a recent prospective randomized phase III clinical trial of patients with metastatic breast cancer (SWOG S0500) showed that early change in therapy based on persistently elevated CTC counts three weeks after starting a drug did not change patient outcome - likely due to poor efficacy of the drugs these metastatic patients received after switching therapy [
36]. One upshot of this study is the expectation that, in the future, the measurement of CTC biomarkers or genotype, rather than only CTC enumeration, should offer better prediction of which drugs will be efficacious.
However, as demonstrated here, primary tumors and metastases can be heterogeneous, and different metastases do not always display the same biological markers. It is not clinically feasible to biopsy all metastases in a given patient at any one time point, and certainly not for serial sampling over the course of disease, so it is hoped that sampling CTCs will reflect the spectrum of tumor cells requiring treatment in metastatic disease. Here in a patient with progressive metastatic breast cancer, we compared the
PIK3CA mutation status of sequentially sampled CTCs to that of tumor cells from two biopsied metastases and DTCs from bone marrow. Unfortunately, and surprisingly, our data did not support the premise that CTCs in most blood draws were reflective of metastatic genotype. While we did show that different metastases contained discordant biomarkers (Table
4),
PIK3CA mutations in CTCs were heterogeneously present in only 2/9 serial blood draws in this patient with multiple distant metastases, two of which (lung and spine) contained tumor cells carrying mutations and whose bone marrow was full of mutant DTCs.
This finding causes pause and suggests that different factors may be at work. One may be that the CTCs analyzed here were captured using the EpCAM cell surface marker. It is postulated that among tumor cells shed from a tissue, many undergo epithelial-mesenchymal transition (EMT), with expression of EMT-associated genes and proteins, and we and others have demonstrated EMT gene and protein expression in CTCs [
21,
37‐
39]. Although CTCs in most EMT studies have been captured with EpCAM antibodies, as EMT progresses, EpCAM expression likely diminishes. Thus, CTCs may consist of populations of EpCAM-expressing and non-EpCAM expressing CTCs. One of the limitations of this study is that because of the technology applied, we studied only EpCAM-expressing CTCs. There may be other non-EpCAM-expressing CTCs present in the blood samples that may have shown a mutant genotype not identified in some of the EpCAM-expressing CTCs. While the mutant tumor cells from metastases in our study were also captured using anti-EpCAM magnetic beads, these cells may potentially have been seeded from EpCAM-negative CTCs that had undergone mesenchymal-epithelial-transition after lodging and growing in the metastatic site, with re-expression of EpCAM on their cell surface. We are now testing different cell surface markers and label-free capture technologies to address this issue, which is particularly important because of recent data suggesting an association between EpCAM-negative CTCs and brain metastases [
40].
A second limitation or explanation of our findings is that we do not know the role of drug treatment in suppressing the appearance of mutant tumor cells in the circulation. For example, at the time that 50 of Patient 12’s CTCs showed no mutation, the patient was receiving RAD001 (everolimus), an mTOR inhibitor that may be more active against cells carrying
PIK3CA mutations [
41]; her CTC count subsequently dropped to zero, perhaps showing response to therapy over time.
A third limitation may be that sequencing only two common hotspots on the
PIK3CA gene may miss other genetic variations that could occur during the evolution of progressive metastatic disease, and which may have been shared by the CTCs and metastases. Very recent and exciting work is underway to develop rigorous methods for investigating single cell whole-exome sequencing of CTCs (also captured using MagSweeper technology) [
42].
However, given current technology development, our study is important in that it adds a note of caution to using CTCs from only a single blood draw to depict the mutational status in any given patient with metastatic disease for treatment purposes. Treatment decisions based on CTC mutational status should only be done under the auspices of a clinical trial.
While CellSearch™ is the only FDA-approved test for enumeration of CTCs, other CTC capture and characterization technologies, including single cell analysis, are rapidly advancing [
43‐
46]. As sequencing technologies progress and cost decreases, it is anticipated that CTC and DTC genotyping will become more clinically feasible. Genotyping of CTCs or DTCs has generally been performed on pooled samples [
47‐
53]. Studying potentially mixed subpopulations, mutant and wild type, may not be as informative regarding which cells respond to which drugs. However, there are reports describing single cell copy number alterations or mutations in CTCs or DTCs from breast [
54‐
56] or other cancers [
12,
57‐
60] using array comparative genomic hybridization and/or sequencing. Like ours, these studies tend to be small, describing only a few cases with aberrant CTC or DTC DNA, but results are encouraging. Tumor cell genotyping at the single cell level may become important clinically because different tumor cell genotypes may be responsive or resistant to different treatments. Thus, capturing and analyzing single tumor cells using deep sequencing of cancer-related genes may lead to even better clarification for selective drug targeting.
However, the lack of mutational discordance between some of the CTC blood draws in Patient 12 and the known presence of mutations in several metastases, raises questions regarding what tumor cells from metastases are released into the circulation, how they may be preferentially impacted by systemic chemotherapy, and whether bone marrow biopsy would be valuable to augment CTC information prior to switching chemotherapy. In the long run, a prospective clinical trial would be needed to determine whether CTCs alone can optimally guide drug therapy and impact survival, or whether assaying both CTCs and DTCs may be better, or whether this is moot until drugs are developed that will ablate metastatic cancer cells.
On a more positive note, we were able to isolate live DTCs and propagate them in culture. Although cell morphology changed over time, mutation status did not. Reports of culturing DTCs are rare [
16,
61], and this is the second to report
in vitro conservation of mutational status [
16]. As extent of growth of DTCs in culture has been previously associated with poor clinical outcome [
16], this lays the groundwork for future
in vitro drug testing experiments using novel therapeutics against DTCs isolated in treatment-refractory patients.
Acknowledgments
The authors thank Ms. Loralee Lobato for her help with clinical coordination and maintenance of the patient database. The authors thank Marc A. Coram, Katharina E. Effenberger, Michael Herrler, and Klaus Pantel for helpful scientific discussions, and Kyra Heirich for assistance during the manuscript revision. This study was supported in part by National Institutes of Health grants R01GM085601 (SSJ), P01HG000205 (RWD) and U54GM62119 (RWD), the John and Marva Warnock Cancer Research Fund, the Andrew and Debra Rachleff Fund, the Hubei 100 Professional Program (GD), the Wuhan 3551 Talents Program (GD), and the Longzhong Talents Plan (GD).
Competing interests
Drs. Stefanie Jeffrey, Ashley Powell, Michael Mindrinos, and Ronald Davis are co-inventors of the MagSweeper device used for tumor cell isolation in this study. Dr Jeffrey has donated her royalties to a non-profit institution. The authors declare that they have no competing interests.
Authors’ contributions
GD, SK, MM, RD, and SJ conceived of the study and participated in its design and coordination. MT contributed patient samples. GD, AP and HZ isolated CTCs, and GD isolated single tumor cells from all other tissues. GD performed the cell culture experiments and immunostains. GD, SK, and MM performed and analyzed the sequencing assays. GD, SK, MM, and SJ drafted the manuscript. All authors have read and approved the final manuscript.