TALK score: Development and validation of a prognostic model for predicting larynx preservation outcome.

OBJECTIVES/HYPOTHESIS To develop and validate a simple prognostic tool that would help predict larynx preservation outcome. STUDY DESIGN A retrospective review of 3 prospective studies. METHODS We reviewed consecutive chemotherapy/radiation protocols for patients (n = 170) with advanced, resectable, squamous cell, larynx, or pharynx cancer treated at Memorial Sloan-Kettering Cancer Center from 1988 to 1995 with larynx preservation intent. The outcome was successful larynx preservation. Model validation used data from U. S. Department of Veterans Affairs larynx preservation study. RESULTS The developed model added one point for each poor prognostic covariate present (show in parentheses) and was given the acronym TALK: T stage (T4), albumin (<4 g/dL), maximum alcohol/liquor use (≥6 drinks/day or heavy drinking), and Karnofsky performance status (<80%). The 3-year larynx preservation rates by TALK score were 65% (0), 41% (1-2), and 6% (3-4), P < .0001; on validation, the TALK 3-4 group was particularly well demarcated. CONCLUSIONS The TALK score is an easily applied and valid tool that should assist treatment selection.