Prognostic modeling of oral cancer by gene profiles and clinicopathological co-variables

Abstract

Accurate staging and outcome prediction is a major problem in clinical management of oral cancer patients, hampering high precision treatment and adjuvant therapy planning. Here, we have built and validated multivariable models that integrate gene signatures with clinical and pathological variables to improve staging and survival prediction of patients with oral squamous cell carcinoma (OSCC). Gene expression profiles from 249 human papillomavirus (HPV)-negative OSCCs were explored to identify a 22-gene lymph node metastasis signature (LNMsig) and a 40-gene overall survival signature (OSsig). To facilitate future clinical implementation and increase performance, these signatures were transferred to quantitative polymerase chain reaction (qPCR) assays and validated in an independent cohort of 125 HPV-negative tumors. When applied in the clinically relevant subgroup of early-stage (cT1-2N0) OSCC, the LNMsig could prevent overtreatment in two-third of the patients. Additionally, the integration of RT-qPCR gene signatures with clinical and pathological variables provided accurate prognostic models for oral cancer, strongly outperforming TNM. Finally, the OSsig gene signature identified a subpopulation of patients, currently considered at low-risk for disease-related survival, who showed an unexpected poor prognosis. These well-validated models will assist in personalizing primary treatment with respect to neck dissection and adjuvant therapies.

Molecular Classification of Lymph Node Metastases Subtypes Predict for Survival in Head and Neck Cancer

“…Previous attempts to biologically classify HNSCCs have mostly relied upon analyses incorporating known patient outcomes to identify mutations and/or differentially expressed genes that correlated with worse survival (25,28,(31)(32)(33). In contrast, fewer studies have used unbiased strategies to classify HNSCCs into distinct biological subtypes that would predict different biological outcomes (26,34,35).…”

Biomarkers: paving stones on the road towards the personalized precision medicine for oral squamous cell carcinoma

“…The validity of this diagnostic microarray was verified by a study on a large number of samples of cN0 OSCC patients, achieving high performance with 86% sensitivity and 89% negative predictive value (NPV) for assessing LNM; this is almost on par with the performance of SLNB [ 109 ]. Recently, the 22-gene LNM signature identified by another study with a large number of samples was applied to pretreatment evaluation of early-stage OSCC patients and effectively reduced the rate of overtreatment by two-thirds [ 110 ].…”

Improved high-dimensional prediction with Random Forests by the use of co-data

“…Data set GSE30788/GSE85446 consists of 150 Dutch patients (of which 60 presented a LNM and 90 did not) with a HPV-negative oral cancer tumor who are in that respect similar to the TCGA patients. Gene expression was measured by microarray, the p -values on GSE30788/ GSE85446 were calculated using a Welch two-sample t-test; further details on the study can be found in [ 29 ]. The differences between the TCGA and the Dutch data (notably the platform and the geographical location of the patients) preclude a straightforward meta-analytic data integration.…”

“…After training the base RF and the CoRF, we validate these classifiers on an independent data set (GSE84846). GSE84846 contains microarray expression data of 97 HPV-negative oral cancer patients from Italy, of whom 49 had a LNM [ 29 ]. To directly apply the classifiers to the validation data, we need to account for the differences in scale between RNASeqv2 and microarray data.…”