Validation of the Risk Model: High-Risk Classification and Tumor Pattern of Invasion Predict Outcome for Patients with Low-Stage Oral Cavity Squamous Cell Carcinoma

Abstract

The risk model is a validated outcome predictor for patients with head and neck squamous cell carcinoma (Brandwein-Gensler et al. in Am j surg pathol 20:167-178, 2005; Am J Surg Pathol 34:676-688, 2010). This model may potentially shift treatment paradigms for patients with low-stage cancers, as current protocols dictate that they might receive only primary surgery. Here we test the hypothesis that the Risk Model has added prognostic value for low-stage oral cavity squamous cell carcinoma (OCSCC) patients. 299 patients with Stage I/II OCSCC were characterized according to the risk model (Brandwein-Gensler et al. in Am J Surg Pathol 20:167-178, 2005; Am J Surg Pathol 34:676-688, 2010). Cumulative incidence and competing risk analysis were performed for locoregional recurrence (LRR) and disease-specific survival (DSS). Receiver operating characteristic analyses were performed for worst pattern of invasion (WPOI) and the risk categories. 292 patients were analyzed; 30 T1N0 patients (17%) and 26 T2N0 patients (23%) developed LRR. Disease-specific mortality occurred in 9 T1N0 patients (6%) and 9 T2N0 patients (10%). On multivariable analysis, the risk model was significantly predictive of LRR (p = 0.0012, HR 2.41, 95% CI 1.42, 4.11) and DSS (p = 0.0005, HR 9.16, 95% CI 2.65, 31.66) adjusted for potential confounders. WPOI alone was also significantly predictive for LRR adjusted for potential confounders with a cut-point of either WPOI-4 (p = 0.0029, HR 3.63, 95% CI 1.56, 8.47) or WPOI-5 (p = 0.0008, HR 2.55, 95% CI 1.48, 4.41) and for DSS (cut point WPOI-5, p = 0.0001, HR 6.34, 95% CI 2.50, 16.09). Given a WPOI-5, the probability of developing locoregional recurrence is 42%. Given a high-risk classification for a combination of features other than WPOI-5, the probability of developing locoregional recurrence is 32%. The Risk Model is the first validated model that is significantly predictive for the important niche group of low-stage OCSCC patients.