In this study, we present a new methodology for using γH2AX as a biomarker to measure drug effects on CTCs from patients during treatment with DNA-damaging agents. γH2AX is a well-established nuclear biomarker for DSB damage (22
). Currently available CTC assay systems are limited to cell membrane antigens, such as HER2, MUC1, and EGFR. In addition, currently used methods for detection of DNA damage in patient samples, including γH2AX immunofluorescence, alkaline comet assay, immunohistochemistry, and FACS analysis can be labor- and time-intensive and have limited applicability in the clinic to monitor tumor response to chemotherapeutic agents. For instance, whereas FACS analysis of γH2AX is more sensitive than the alkaline comet assay (41
), it does not distinguish CTCs from normal blood cells, and the analytical sensitivity is too low to detect CTCs, even if it could distinguish them; therefore, γH2AX response to drug treatment could be established only in non-tumor cells (e.g., peripheral blood mononuclear cells). Compared with other approaches, the combination of CTC enrichment with nuclear γH2AX detection is not only a unique and advanced technique, but also feasible for incorporation into clinical trials of molecular-targeted oncology drugs.
Validation studies, including CTC recovery, assay accuracy, and assay precision, demonstrate that this is a reproducible, reliable, and quantitative method (Supplementary Material
). As with previous studies (19
), minimal variation in detection of CTCs from duplicate samples is observed. In addition, assay variation in detection of γH2AX-positive CTCs induced with therapeutically relevant concentrations of topotecan was less than 10%.
Studies employing cancer cell lines treated with increasing doses of topotecan, then spiked into whole blood, showed that the fraction of γH2AX-positive cells increased in a dose-dependent manner, and treatment of cancer cell lines with rapamycin showed the specificity of the γH2AX response for DNA damage. In addition, our validation studies determined that there is an expected high variation and low γH2AX-positive cell recovery when initial CTC recovery is low, but minimal day-to-day variation in CTC recovery (less than 20%) when CTC counts were above 10 cells per sample. For those patients with low to no CTCs, it may be necessary to find alternate capture antibodies for evaluation.
Using this validated assay, we performed a population baselining study to assess the range of CTC counts in our patients and the day-to-day variability in CTC counts in single patients. Initial CTC counts were determined for 15 patients with a variety of cancer types; 73% (11 of 15) of the patients had measureable CTCs in their blood prior to treatment. We also determined that following CTC enrichment with the CellSearch system, staining with the γH2AX-AF488 antibody yielded a clear nuclear signal, and that the fraction of γH2AX-positive CTCs was low and consistent in patients not undergoing radiation treatment or chemotherapy.
We then evaluated patients on various chemotherapy regimens and found that the fraction of γH2AX-positive CTCs increased in patients treated with topotecan alone, a combination of topotecan and a PARP inhibitor, or a combination of cyclophosphamide and PARP inhibitor. Increased numbers of γH2AX-positive cells could be measured irrespective of increases or decreases in the total number of recovered CTCs. The change in the number of CTCs induced by drug treatment varied from patient to patient and may be dependent on individual short-term drug response. In the case of patients with high baseline CTC counts, CTC counts decreased following treatment. However, for patients with zero or low baseline CTCs, CTC counts increased post-treatment. In this second group of patients, it appears that CTCs were shed from the tumor in response to therapy. If this interpretation is correct, the γH2AX response detected in CTCs was not likely due to CTCs already in the peripheral blood being exposed to the drug, but instead directly reflected the drug-induced DNA damage in cells in the tumor.
Meng and colleagues estimated CTC half-life by flow cytometry using specimens from five breast cancer patients pre- and peri-surgery (43
). All five patients had declining numbers of CTCs post-surgery, with 6 of 10 aggregated data points falling into the background noise level of the instrument within the first 24 h. Their conclusion of a CTC half-life measured in hours is neither supported nor contradicted by our data, since the majority of patients in our study had increased CTC numbers post-chemotherapy. However, if their conclusion is correct, the half-life estimate would strengthen our argument that γH2AX-positive CTCs collected 24 h after a single administration of drug are recently shed from the tumor and reflect a pharmacodynamic response of the tumor. It should be noted that the patients in our study all have active disease (with distant metastases) that has failed numerous rounds of therapy. Thus, our current data are limited to patients in phase I trials, which may not be a fair representation of the overall patient population in oncology. It is possible that patients with disease that has not responded to multiple lines of therapy have CTCs with different cell surface markers. We are exploring whether markers other than EpCam will isolate more CTCs in patients at the National Cancer Institute. It remains to be established, however, whether the fraction of γH2AX-positive CTCs is representative of the positive fraction in the tumor, and this will be investigated in upcoming clinical trials. In addition, the half-life of CTCs in peripheral blood is not known and needs to be determined once a suitable method has been developed.
Based on our current data, several additional questions and limitations need to be considered. First, the current assay is limited to chemotherapeutic agents that induce a DNA damage response through the ATM/γH2AX pathway; however, alternate nuclear biomarkers could be used to expand this approach (11
). Second, while the number of isolated CTCs is currently used as a prognostic indicator for disease outcome in patients with breast and colon cancer, the utility of monitoring changes in the number of γH2AX-positive CTCs during treatment, and the types of cancers this technology can be reliably employed to monitor are yet to be determined and will require broader testing. Finally, it is unknown whether γH2AX in CTCs will correlate with clinical efficacy (tumor shrinkage). To address these last points, this assay is currently being incorporated as a correlative study in early-phase clinical trials at the National Cancer Institute, and drug effect on disease progression will be correlated to γH2AX-positive CTCs.
In conclusion, γH2AX is a sensitive biomarker for monitoring the pharmacodynamic effects of topotecan alone, a combination of topotecan and PARP inhibitor, or a combination of PARP inhibitor and cyclophosphamide in isolated CTCs. This method only requires patient blood samples, allowing the possibility of monitoring drug effect throughout treatment. This method has obvious advantages over invasive tumor biopsies, as well as tumor imaging, which often requires a significant span of time between images to detect effects from treatment. In combination with CTC counts, monitoring of γH2AX signal could lead to enhanced monitoring of chemotherapeutic effects in the clinic. In ongoing clinical trials within the National Cancer Institute, we are testing whether assaying γH2AX levels in CTCs provides a more sensitive marker for assessing pharmacodynamic effects in patients than total CTC counts alone. Another potential application of this method is to determine the half-life of CTCs after shedding from the tumor; such measurements could have both theoretical and diagnostic significance. Finally, the strategy employed for our method can be adapted for additional nuclear biomarkers in CTCs. We hope this will become a useful tool to evaluate pharmacodynamic effects in clinical trials of novel DNA-damaging agents.