OBJECTIVE: Pancreatic cancers are characterized by dense stroma. To estimate the degree of interference by coexisting noncancer cells in molecular analyses, we aimed to develop a DNA methylation marker that assesses a cancer cell fraction in DNA samples. METHODS: The microarray data of 22 pancreatic cancer tissues from the The Cancer Genome Atlas database and 9 noncancer tissues were used for genome-wide screening. Thirty-one surgical tumor samples (10 intraductal papillary mucinous neoplasms [IPMNs] and 21 pancreatic cancers), 4 normal, and 26 nontumor samples were used for validation. Gene-specific methylation analysis was conducted by bisulfite pyrosequencing. RESULTS: Genome-wide screening isolated SIM1, MIR129-2, NR1I2, and HOXB-AS4, as specifically methylated in pancreatic cancer cells. Bisulfite pyrosequencing validated that one or more of three genes (SIM1, MIR129-2, and NR1I2) were methylated in 22 (71.0%) tumor samples (8 IPMNs and 14 cancers), and all showed low levels of methylation in 26 (86.7%) normal and nontumor samples. Therefore, the three genes collectively constituted one marker for a pancreatic cancer cell fraction. The cancer cell fraction estimated by the marker was highly correlated with that estimated using the KRAS mutant allele frequency (R = 0.79). CONCLUSION: The DNA methylation marker is useful to estimate the pancreatic cancer cell fraction in DNA samples.
OBJECTIVE:Pancreatic cancers are characterized by dense stroma. To estimate the degree of interference by coexisting noncancer cells in molecular analyses, we aimed to develop a DNA methylation marker that assesses a cancer cell fraction in DNA samples. METHODS: The microarray data of 22 pancreatic cancer tissues from the The Cancer Genome Atlas database and 9 noncancer tissues were used for genome-wide screening. Thirty-one surgical tumor samples (10 intraductal papillary mucinous neoplasms [IPMNs] and 21 pancreatic cancers), 4 normal, and 26 nontumor samples were used for validation. Gene-specific methylation analysis was conducted by bisulfite pyrosequencing. RESULTS: Genome-wide screening isolated SIM1, MIR129-2, NR1I2, and HOXB-AS4, as specifically methylated in pancreatic cancer cells. Bisulfite pyrosequencing validated that one or more of three genes (SIM1, MIR129-2, and NR1I2) were methylated in 22 (71.0%) tumor samples (8 IPMNs and 14 cancers), and all showed low levels of methylation in 26 (86.7%) normal and nontumor samples. Therefore, the three genes collectively constituted one marker for a pancreatic cancer cell fraction. The cancer cell fraction estimated by the marker was highly correlated with that estimated using the KRAS mutant allele frequency (R = 0.79). CONCLUSION: The DNA methylation marker is useful to estimate the pancreatic cancer cell fraction in DNA samples.