In some cases of early-stage breast cancer, the seeds of metastasis have already been planted even when treatment seems successful. A method for predicting which women will relapse could help in designing clinical trials for more effective treatments. A team led by Nicholas C. Turner of Royal Marsden Hospital and the Institute of Cancer Research, both in London, has found that it can predict relapse in early-stage breast cancer by tracking mutations in circulating tumor DNA (ctDNA) (Sci. Transl. Med. 2015, DOI: 10.1126/scitranslmed.aab0021). The researchers used digital polymerase chain reaction assays to quantify specific mutations in ctDNA in plasma samples from 55 women. For each patient, they obtained plasma samples before and after surgery and at six-month intervals for two years or until relapse. They also collected samples from most patients two to four weeks after surgery and used them to determine the benefit of analysis at a single time point. Serial sampling was better for predicting relapse than single-point analysis. The study shows that measurement of ctDNA can predict relapse months before it is detectable with conventional scans.