5562 Background: TNM staging is an effective tool for prognostic prediction in patients with oral cavity cancer, but it disregards variables such as the anatomic subsite, medical comorbidity, lifestyle, etc. A nomogram has the ability to take into account many factors predictive of outcome for an individual patient, beyond the traditional TNM. We have developed a nomogram to accurately predict overall mortality (OM) and disease specific mortality (DSM) in individual patients treated for oral cancer. Methods: Demographic, host, and tumor characteristics of 1,617 oral cancer patients treated with surgery and adjuvant treatment between 1985 and 2009 were available from a preexisting database. Recurrent disease was recorded in 509 patients, 328 died of cancer-related causes and 542 died of other causes. The median follow-up was 42 months (range 1-300 months). Cox proportional hazards regression model was used for OM and competing risk regression was used for DSM. Death from other causes was treated as competing risk for DSM. Missing values in the predictors were multiply imputed before analysis. Variables analyzed as prognosis predictors included age, gender, race, alcohol/tobacco use, anatomic subsite, comorbidity, tumor size/thickness, bone/deep muscle invasion, nodal status/level, grade, vascular/perineural invasion, margin status and adjuvant therapy. Step-down method was used to select the statistically most powerful predictors for inclusion in the final nomogram for each outcome. Results: Age, severe comorbidity, subsite, tumor size, bone/deep muscle invasion, margin status, vascular and perineural invasion, nodal status and level, and adjuvant therapy were the variables with highest predictive value for OM. The most influential predictors of DSM were gender, tumor size, nodal status and location, subsite, margin status, grade and vascular invasion. Nomograms were generated for prediction of OM and DSM. The nomograms were internally validated with correction for over-fitting; bias-corrected concordance index for OM was 75% and DSM 76%. Conclusions: We have developed nomograms that can accurately estimate OM and DSM based on tumor and host characteristics of an individual patient treated for oral cancer.
Clinicopathological Profile & Correlation of Tumor Size with Nodal Status in Oral Cavity Cancers: An AHRCC Experience
