Suraj Peri , 1 , Evgeny Izumchenko 2 , Adrian D. Schubert 2 , 6 , Michael J. Slifker 1 , Karen Ruth 1 , Ilya G. Serebriiskii 3 , Theresa Guo 2 , 4 , Barbara A. Burtness 5 , Ranee Mehra 2 , 4 , Eric A. Ross 1 , David Sidransky 2 , 4 , Erica A. Golemis , 3
24 November 2017
Squamous cell carcinomas of the head and neck (SCCHN) affect anatomical sites including the oral cavity, nasal cavity, pharynx, and larynx. Laryngeal cancers are characterized by high recurrence and poor overall survival, and currently lack robust molecular prognostic biomarkers for treatment stratification. Using an algorithm for integrative clustering that simultaneously assesses gene expression, somatic mutation, copy number variation, and methylation, we for the first time identify laryngeal cancer subtypes with distinct prognostic outcomes, and differing from the non-prognostic laryngeal subclasses reported by The Cancer Genome Atlas (TCGA). Although most common laryngeal gene mutations are found in both subclasses, better prognosis is strongly associated with damaging mutations of the methyltransferases NSD1 and NSD2, with findings confirmed in an independent validation cohort consisting of 63 laryngeal cancer patients. Intriguingly, NSD1/2 mutations are not prognostic for nonlaryngeal SCCHN. These results provide an immediately useful clinical metric for patient stratification and prognostication.
The authors use an integrative clustering approach to identify two laryngeal cancer clusters with distinct prognosis and show that mutations damaging the NSD1 and NSD2 methyltransferases segregate to the cluster with favorable prognosis, and independently predict longer survival in patients with laryngeal, but not other head and neck cancers.