Florent Mouliere and colleagues have designed a new method that was able to detect hard-to-trace tumor DNA in the blood (or ctDNA). Their technique could outperform the cancer-sniffing capabilities of existing methods by accounting for differences in the size of ctDNA fragments, rather than genomic changes in ctDNA. The blood plasma of cancer patients contains DNA that originates from primary or secondary tumors, and researchers have suspected that zeroing in on ctDNA may offer a safe and noninvasive alternative compared to conventional biopsies - requiring only a small blood sample. However, ctDNA is difficult to identify because it is typically outnumbered by much larger quantities of noncancerous DNA that also circulates in the blood. Previous research has shown that fragments of circulating DNA from fetuses can be distinguished from maternal DNA because of their shorter lengths, a finding that helped to improve the sensitivity of prenatal diagnosis. Taking inspiration from this discovery, Mouliere et al. used whole-genome sequencing to survey the sizes of ctDNA fragments in 344 plasma samples from 200 cancer patients. They identified biological features that were enriched in ctDNA associated with each patient's cancer type, and developed computational methods to select for specific ctDNA fragment sizes. The authors found that focusing on fragments between 90 and 150 base pairs long - the unit of measurement for genetic material - improved detection of ctDNA from patients with brain, renal and pancreatic cancer. The authors say that size selection could have an impact on the detection of other types of DNA (such as mitochondrial DNA) in body fluids. A related Focus by Ellen Heitzer and Michael. R. Speicher describes the study in more detail.