Andy Trotti

5552 Background: Epithelial-to-mesenchymal transition (EMT) in patients with head and cancer (HNC) has been associated with high risk of recurrence and poor survival. To identify a clinically applicable biomarker of EMT, we examined the expression levels of E-cadherin (E-cad), beta-catenin (B-cat), and glioma-associated oncogene homolog 1 (Gli1). They were also examined in context of epidermal growth factor receptor (EGFR) expression. Methods: Cytoplasmic E-cad, nuclear B-cat, and nuclear Gli1 were determined in tumors from patients enrolled in RTOG 9003 (radiation fractionation trial) using automated quantitative analysis. The results were also correlated with previously determined EGFR expression. The expression levels were evaluated in relation to 3 endpoints: time to metastasis (TTM), time to disease progression (TDP), and overall survival (OS). Results: Among 1,068 eligible patients, E-cad, B-cat, Gli1, and EGFR data were available in 326, 236, 339, and 300 patients, respectively. Expression of pairs of E-cad, B-cat, and Gli1 were significantly correlated, but not with expression of EGFR. While E-cad and B-cat expression did not differentiate prognosis, Gli1 was associated with poorer outcomes, adjusted for age, T and N stages, and Karnofsky performance status, and the significant influence persisted in a multivariable analysis. Conclusions: These data suggest that Gli1 may play an important role in HNSCC prognosis, especially in TTM, but its role in EMT process needs further investigation. In addition, risk classification based on this analysis is exploratory and independent validation on other cohorts is necessary. Supported by RTOG U10 CA21661 and CCOP U10 CA37422 grants from the NCI. This abstract's contents are the sole responsibility of the authors and do not necessarily represent the official views of the NCI. No significant financial relationships to disclose.

6026 Background: There is currently no method for concisely summarizing and comparing aggregate adverse event (AE) data among treatment options. We propose a new concept (“toxicity burden”) and new metrics to reflect treatment-related risk. Analogous to the T-N-M tumor staging system, the “T-A-M End-results system” (TAME) consolidates adverse effects data into a concise index: short-term (acute) toxicity [T], adverse long-term effects [A] and mortality risk from treatment [M]. Methods: We estimated T-scores and A-scores from published adverse event data in 13 treatment arms from 5 RTOG H&N trials (90–03, 91–11, 95–01, 97–03, 99–14) conducted between 1990–2000 (>2200 patients). Estimation of T and A scores are calculated by dividing total number of RTOG grade 3–4 events for 13 pre-specified toxic categories by the number of patients in a cohort. We also calculated scores using individual patient data (IPD) from 2 arms of the RTOG 91–11 larynx preservation trial (314 patients). Results: Estimated T-scores demonstrate more than a 500% difference in acute toxicity burden when comparing aggressive chemoradiotherapy to conventional radiotherapy (range: 0.47–3.14). Less variation was seen in estimated late effects burden (range: 0.17–0.59). Putative T-risk burden classes were designated: ≤0.6 (low), 0.7–1.0 (mod), 1.0–2.5 (high), ≥2.5 (extreme). T-values from RTOG 91–11 (see table) suggest up to 250% more toxicity burden from the use of chemotherapy. Distribution of T-scores among the concurrent PLAT-RT group revealed 22% of patients experienced 3 high-grade events, and 23% experienced 4 or more (p=<0.0001). Correlations of T-scores with KPS and selected QOL domains at 3 months were modest as expected. Conclusions: TAME scores support the feasibility of quantifying toxicity burden and may have wide applications including safety monitoring and identification of patient groups at high risk for injury in multiple oncologic disorders. No significant financial relationships to disclose.