Details

Autor(en) / Beteiligte

• Seitz, Rob
• Hout, David Richard
• Ring, Brian Z
• Bailey, Dan B

Titel

Association of mesenchymal subtype with the presence of immune infiltrates in squamous cancer classifier

Ist Teil von

• Journal of clinical oncology, 2017-05, Vol.35 (15_suppl), p.e23073-e23073

Erscheinungsjahr

2017

Quelle

Free E-Journal (出版社公開部分のみ）

Beschreibungen/Notizen

• Abstract only e23073 Background: Lehmann and colleagues (Lehmann 2011) devised a classification system for triple negative breast cancer (TNBC) consisting of seven subtypes—IM, BL1, BL2, LAR, M, MSL, and UNS. Recent work has shown the IM group (characterized by the presence of immunomodulatory genes) is due to tumor infiltrating lymphocytes (TILs), and should be measured independently of the other subtypes. Furthermore, the m subtype (characterized by genes associated with epithelial-mesenchymal transition (EMT)) was shown to have a highly inverse relationship with the IM group (Lehmann 2016). We (Ring 2016) modified the original algorithm into a 101-gene algorithm that also confirmed this inverse relationship between IM and m signatures (Grigoriadis SABCS 2016). While this is a novel and interesting finding, we wanted to know whether this inverse relationship applies to other epithelial cell origin cancers such as squamous lung cancer. Methods: We downloaded gene expression data from 548 squamous cell lung cancers from the TCGA database, and applied the 101-gene TNBC algorithm to the lung cohort without optimization or adjustment. Results: Of the 548 samples, 389 (71%) were able to be classified by the 101-gene algorithm. Of these, 82 (21%) had an m phenotype. 128 patients (33%) were positive for the IM signature. The expected overlap of patients positive for both IM and m was 27 patients (7%). The observed overlap was 0 patients (chi-squared = 29; p < 0.001). Conclusions: The unoptimized 101-gene TNBC algorithm validated the observation that gene expression signatures for IM and m are inversely correlated. While any attempt to apply a TNBC algorithm to squamous lung cancer would need careful optimization and validation, these data support that the inverse relationship noted in TNBC may transcend the tissue of origin.