163 Detection of non-metastatic non-small cell lung cancer in urine independent of smoking status (17, 19), although smoking status was defined differently. To further rule out bias due to smoking status, more smoking controls should be included in future studies. Furthermore, to determine whether the urine assay is able to differentiate between lung cancer and other pulmonary diseases, future studies should also include patients with non-malignant pulmonary pathology. Nevertheless, from a screening perspective, subjects without any (pulmonary) medical history will represent a large proportion of the screening population. A second potential limitation of the present study is the use of only one urine sample per patient as previous work found that DNA methylation levels might vary greatly over time, both between- and within advanced-stage NSCLC patients (25). A key strength of the current study is that study participants collected urine from home and sent it to the laboratory by regular mail. This is an appropriate setting to evaluate the use of a self-collected specimen to make screening more accessible. Another strength is the measurement of three methylation markers and a reference gene within a single PCR reaction to reduce costs, time, and the amount of input DNA. Although numbers were small, the collection of a second urine sample after surgery of a subset of patients allowed exploring whether methylation levels altered after surgery with curative intent. Several technical improvements can be suggested for future studies. In this study, a commercial kit for column-based DNA extraction was used. However, other protocols that enable isolation of short fragmented urine cfDNA, such as methylation on beads (17, 19), hybridization capture, and Q Sepharose DNA isolation (45), might prove superior. Furthermore, bisulfite-free procedures using modified sequencing techniques or nanotechnology-based electrochemical biosensors might facilitate a more sensitive and robust detection of DNA methylation (46-48). Due to the dynamic nature of cfDNA in the urine of NSCLC patients, collecting multiple urine samples per patient could also increase the accuracy of NSCLC detection in urine (25). A genome-wide screen across gene promoter regions using urinary cfDNA of non-metastatic NSCLC patients may yield more accurate biomarkers applicable to urine samples. The combination of DNA methylation with other ctDNA aberrations, such as mutations, copy number alterations or differences in fragment lengths (49), but also non-DNA tumor derivatives in urine such as proteomics or metabolomics, might further improve the performance of urinebased cancer tests (50, 51). In conclusion, the present study demonstrates technical feasibility of detecting nonmetastatic NSCLC in urine using ctDNA methylation analysis. Further research including more patients is needed to validate this approach. 6
RkJQdWJsaXNoZXIy MjY0ODMw