A computational workflow translates a 58-gene signature to a formalin-fixed, paraffin-embedded sample-based companion diagnostic for personalized treatment of the BRAF-mutation-like subtype of colorectal cancers

Sjors G.J.G. In’t Veld, Kim N. Duong, Mireille Snel, Anke Witteveen, Inès J. Beumer, Leonie J.M.J. Delahaye, Diederik Wehkamp, René Bernards, Annuska M. Glas, Sun Tian*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Colorectal cancer patients with the BRAF(p.V600E) mutation have poor prognosis in metastatic setting. Personalized treatment options and companion diagnostics are needed to better treat these patients. Previously, we developed a 58-gene signature to characterize the distinct gene expression pattern of BRAF-mutation-like subtype (accuracy 91.1%). Further experiments repurposed drug Vinorelbine as specifically lethal to this BRAF-mutation-like subtype. The aim of this study is to translate this 58-gene signature from a research setting to a robust companion diagnostic that can use formalin-fixed, paraffin-embedded (FFPE) samples to select patients with the BRAF-mutation-like subtype. BRAF mutation and gene expression data of 302 FFPE samples were measured (mutants = 57, wild-type = 245). The performance of the 58-gene signature in FFPE samples showed a high sensitivity of 89.5%. In the identified BRAF-mutation-like subtype group, 50% of tumours were known BRAF mutants, and 50% were BRAF wild-type. The stability of the 58-gene signature in FFPE samples was evaluated by two control samples over 40 independent experiments. The standard deviations (SD) were within the predefined criteria (control 1: SD = 0.091, SD/Range = 3.0%; control 2: SD = 0.169, SD/Range = 5.5%). The fresh frozen version and translated FFPE version of this 58-gene signature were compared using 170 paired fresh frozen and FFPE samples and the result showed high consistency (agreement = 99.3%). In conclusion, we translated this 58-gene signature to a robust companion diagnostic that can use FFPE samples.

Original languageEnglish
Article number16
Issue number4
Publication statusPublished - 1 Dec 2017
Externally publishedYes

Cite this