Does Operability Status Influence Outcomes in Patients with T4 Larynx Cancer Undergoing Larynx Preservation?

Abstract

<h3>Purpose/Objective(s)</h3> Analyses of national care patterns and outcomes among patients with T4 larynx cancer (LC) have raised concerns that inappropriate utilization of larynx preservation (LP) has contributed to reduced survival compared to laryngectomy (LGX). Little consideration has been given to confounding by indication, wherein patients ineligible for standard of care receive non-standard therapy. We tested the hypothesis that operability status is an independent predictor of survival in patients with T4 LC and that patients who are deemed operable and who undergo LP do not have worse overall survival than those undergoing LGX. <h3>Materials/Methods</h3> We queried the National Cancer Database for cases of T4M0 LC diagnosed from 2004-2015. Patients were categorized as undergoing either LGX, LP with concurrent chemoradiotherapy (CRT) when coded as being offered surgery but declining it (LP-operable), and LP undergoing CRT when lacking these codes (LP-inoperable). Overall survival (OS) was estimated by Kaplan-Meier. Cox univariable (UVA) and multivariable analyses (MVA) were used to identify variables associated with OS. Age, sex, race, year of diagnosis, primary payor, facility type, Charlson-Deyo score, primary site, cT-subclassification, cN-classification, and treatment/operability group were entered as variables in the model. Analysis was repeated in a subset identified using inverse probability of treatment weights (IPTW) derived from propensity scores. <h3>Results</h3> We identified 1,405 LGX, 164 LP-operable and 1,969 LP-inoperable patients, with median OS of 43.7, 40.3, and 31.4 months, respectively (<i>p</i><0.01). Compared to LGX patients, Cox MVA demonstrated significantly worse OS among LP-inoperable patients (HR 1.25, 95%CI 1.14-1.37, <i>p</i><0.001) but not among LP-operable patients (HR 1.12 95%CI 0.91-1.38, <i>p</i>=0.288). These findings were maintained in the IPTW sample: Median OS was 42.0, 40.3, and 31.1 months for LGX, LP-operable, and LP-inoperable patients, respectively (<i>p</i><0.01). Compared to LGX patients, Cox MVA revealed worse OS among LP-inoperable patients (HR 1.21 95%CI 1.11-1.33, <i>p</i><0.001) but not LP-operable patients (HR 1.03 95%CI 0.84-1.28, <i>p</i>=0.754). <h3>Conclusion</h3> Patients with T4 LC who are deemed operable but undergo LP do not have significantly worse OS compared to those undergoing LGX. Our findings suggest that prior retrospective comparisons have been confounded by operability status and indicate that there may be a cohort of patients with T4 disease who may be appropriately treated with LP. Additional research is needed to determine which T4 patients may be eligible for LP.