Measuring Circulating Tumor Cells As a Surrogate End Point for Adjuvant Therapy of Breast Cancer: What Do They Mean and What Should We Do About Them?

The biologic role of dislodged cancer cells that are circulating in the bloodstream and the value of detecting them has long been a subject for speculation. The fear of inducing such shedding from the primary tumor site was the basis of frequently taught principles of surgical technique. In this issue of the Journal of Clinical Oncology, Pachmann et al describe a modified assay of breast-cancer–circulating tumor cells (CTC, or circulating epithelial tumor cells [CETC] in their nomenclature), which is sensitive enough to detect tumor cells in the peripheral blood of most postoperative patients with breast cancer who otherwise have no detectable disease. Whenever we intend to treat cancer patients with potentially toxic therapies, it would be ideal to know in advance whether tumor burden and our ability to reduce it will produce a clinical benefit. This concept has been investigated extensively in breast cancer with modest success, but few if any absolute predictors of benefit have emerged. Many indicators of prognosis have been described, ranging from “simple” markers such as hormone receptor status, nuclear grade, tumor size, nodal status, and plasminogen activator, to more complex gene expression profiles. Others have focused on finding microscopic deposits of tumor cells in distant sites or in the circulation. These markers certainly inform us about the likelihood of recurrence, but most do not indicate the likelihood that treatment will reduce this risk. Some markers are both prognostic and predictive, identifying tumors that are more likely to recur and also those that are more likely to be affected by treatment. The OncotypeDx recurrence score is a good example of such a “mixed” marker, and its potential impact on patient outcomes is now being explored in a large prospective trial. However, most of these markers and profiles really apply to populations of patients; with rare exceptions (such as the absence of hormone receptor expression), none will tell us with any degree of certainty that a particular treatment will be either useful or futile in an individual patient. Even when a patient survives long-term without recurrence, we cannot know whether this represents success of treatment or the initial absence of micrometastases. Being able to predict the response to therapy has been more closely approximated in the neoadjuvant setting, which is one of the main strengths of this approach. With neoadjuvant treatment, the accuracy of predictive markers can be determined by direct measurement of the treatment’s effect on the tumor. Even in this setting, however, profiles that correlate statistically with pathologic complete response (pCR), do not tell us whether an individual patient’s tumor will respond in a meaningful way. pCRs occur in only a minority of patients, but most patients experience objective responses and may still benefit clinically even if the tumor does not disappear completely. As compared to the ideal of being able to determine ahead of time whether a specific treatment will benefit an individual patient, a surrogate end point would be able to tell us whether a therapy that is being delivered is affecting the true clinical target. In patients with measurable metastatic disease or in the setting of neoadjuvant therapy, the response of tumor that can be palpated or measured on imaging allows us to judge the response to treatment in progress. Even if the primary tumor response is not a perfect surrogate for patient outcomes, it is an excellent predictor of disease-free survival and overall survival. For patients with unmeasurable metastatic disease, the benefit of palliative chemotherapy or hormonal therapy is often determined based on symptoms or serum markers. A well-validated assay of CTC has recently been described, and appears to be a good predictor of outcomes in patients with metastatic breast cancer, when measured either before therapy or at follow-up after treatment. But despite the potential utility of this assay in patients with metastatic disease, its low sensitivity makes it unhelpful in the adjuvant setting. In contrast, by eliminating purification steps that also remove tumor cells, the assay described in this issue by Pachmann et al detected CTCs in 90% of their adjuvant breast cancer patients. It should be kept in mind, of course, that all of the patients selected for this study were candidates for adjuvant chemotherapy based on classical risk factors. Most importantly, the pattern of change in CTC during therapy was highly predictive of patient outcomes. Three patterns of response were detected: 10-fold or more decrease in CTC number, marginal change in CTC, or greater than 10-fold increase in CTC. The last group sometimes demonstrated an increase after an initial decrease in CTC numbers. Only one relapse was observed in the first group, while the majority of recurrences were seen in the third group. But is this a truly predictive measure that tells us whether the therapy is working or is it just another prognostic marker?