Development of hemodiagnosis for early detection of ovarian cancer using glycopeptide peaks obtained from CSGSA (Comprehensive Serum Glycopeptide Spectra Analysis)

2020 Fiscal Year Final Research Report

• Project/Area Number: 17H04340
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Obstetrics and gynecology
• Research Institution: Tokai University
• Principal Investigator: MIKAMI Mikio 東海大学, 医学部, 教授 (30190606)
• Co-Investigator(Kenkyū-buntansha): 信田 政子 東海大学, 医学部, 講師 (10338717) ; 池田 仁惠 東海大学, 医学部, 講師 (20365993) ; 柴田 健雄 東海大学, 健康学部, 講師 (30366033) ; 宮澤 昌樹 東海大学, 医学部, 客員講師 (30624572) ; 平澤 猛 東海大学, 医学部, 准教授 (70307289)
• Project Period (FY): 2017-04-01 – 2021-03-31
• Keywords: 卵巣癌 / 血清バイオマーカー / 糖ペプチド / 質量分析 / 人工知能 / 深層学習 / リキッドバイオプシー

Outline of Final Research Achievements

Ovarian cancer is a leading cause of deaths among gynecological cancers, and a method to detect early-stage epithelial ovarian cancer (EOC) is urgently needed. FS-C4BP, which was discovered by CSGSA, can be measured by Lectin and immunoblot to detect early EOC. CSGSA evaluates >10,000 glycopeptides and identifies reproducible peaks and patterns via supervised orthogonal partial least-squares discriminant modeling（OPLS-DA）. Combined CSGSA（OPLS-DA）, CA125, and HE4 had improved diagnostic performance. We also developed an AI-based CSGSA method. We converted serum glycopeptide expression patterns into two-dimensional (2D) barcodes to let convolutional neural network (CNN) learn and distinguish between EOC and non-EOC. When CNN was trained with 2D barcodes colored on the basis of serum levels of CA125 and HE4, a diagnostic AUC of 95% was achieved. This simple and low-cost method will increase the detection of EOC. Thus, CSGSA may be a useful screening tool for detecting early stage EOC.

Free Research Field

婦人科腫瘍学、糖鎖生物学

Academic Significance and Societal Importance of the Research Achievements

卵巣癌は早期発見が難しくかつ予後も極めて悪い癌であり、新たな発想の新規診断技術導入が重要である。腫瘍マーカーは単一分子と認識され研究されてきたが、現状では卵巣癌早期診断は不可能であろう。そこで古い概念を打ち破り、究極のCombination Assayと考えられる網羅的血清糖ペプチドピークと人工知能を用いた卵巣癌早期診断を開発し、現在汎用されている卵巣癌マーカーであるCA125とHE4よりも有意に初期卵巣癌を判別できる診断法を開発した。