Purpose: Cancer of unknown primary site (CUPS) is a group of cancers for which the anatomic site of origin remains occult after detailed investigation. Until now, immunohistochemistry and tissue-specific RNA expression pattern have been used to predict tissue-of-origin in CUPS. However, these techniques do not fulfill the sensitivity and specificity for clinical practice. Recently, several studies found the tissue-specific methylation patterns in the genome. In this context, we tried to discover the tissue-specific methylation markers by analyzing genome-wide methylation data in The Cancer Genome Atlas (TCGA). Then, we aimed to develop the methylation-specific next-generation sequencing panel that predicts the tissue-of-origin in CUPS by validating panels in clinical samples.

Experimental Design: We selected every 17 mostly hypermethylated CpG sites in 31 cancer types by analyzing 8,350 cases of Infinium 450K methylation data in TCGA. With the selected 527 CpG sites as input variables, we constructed 465 decision tree models for all pairwise classification of 31 cancer types by using C50 package in R. Based on the two class classification models, we implemented an ensemble voting classifier for tissue-of-origin prediction test in fresh frozen tissue of 50 primary cancers of 8 tissue types using methylation-specific next-generation sequencing, which targeted 527 tissue-specific CpG sites by post-bisulfite method.

Results: In in silico analysis, we randomly selected 80% of the TCGA data for training decision tree models and used the remaining 20% of the data for testing those models. The sensitivity and specificity for the tissue-of-origin in 31 cancer types of testing data were 0.91 and 0.99 in decision tree models, respectively. In methylation-specific next-generation sequencing, our model predicted 40 cases (80%) of total 50 clinical samples, successfully.

Conclusions: Prediction of tissue-of-origin using methylation-specific next-generation sequencing might be promising. For clinical application, further study using more comprehensive targeted resequencing panel using formalin-fixed, paraffin-embedded tissues will be followed.

Citation Format: Jeong Mo Bae, Kwangsoo Kim, Hee Jun Chae, Xianyu Wen, Kang Yeol Kim, Hwang Kwan Gwon, Nam-Yun Cho, Gyeong Hoon Kang. Identification of tissue-of-origin in cancer of unknown primary site (CUPS) using methylation-specific targeted resequencing: A pilot study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3312.