Computed tomography evaluation after induction chemotherapy for T3 laryngeal cancer: Does response correlate with vocal cord mobility?

Publication date: March 2019

Source: Oral Oncology, Volume 90

Author(s): Gabriel C.T.E. Garcia, Philippe Gorphe, Dana Hartl, Samy Ammari, Caroline Even, Yungan Tao, Corinne Balleyguier, Arthur Varoquaux, François Bidault

Abstract

Objectives

After induction chemotherapy (IC) for laryngeal cancer, Computed Tomography (CT) is used to assess tumor response but lacks rationalized methods for measurement of this response. In T3 laryngeal cancer, remobilization of an initially fixed vocal cord (VC) is a major sign of tumor response. We compared the performances of RECIST1.1, WHO and volumetric methods of evaluating response with laryngeal remobilization to determine if these measurements could objectively assess tumor response.

Material and Methods

This monocentric retrospective cohort study included patients treated with T3 laryngeal cancer with initial VC fixation treated with an organ preservation protocol with IC between 1999 and 2012. Tumors were measured with CT using RECIST1.1, WHO and volumetric methods by two radiologists blinded to VC remobilization (VCR), before and after induction chemotherapy and VC mobility was clinically assessed. Radiologic tumor shrinkage was compared to VCR. AUCs of ROC curves were compared. Inter-reader reliability, sensitivity and specificity of optimal cutoffs were determined.

Results

Seven females and thirty-six males with a mean age of 59 years were included. AUCs were 0.759, [95%CI 0.614; 0.904] for RECIST1.1, 0.763, [95%CI 0.617; 0.909] for WHO and 0.752, [95%CI 0.608; 0.896] for volumetric evaluations with no significant difference among the three techniques. Inter-reader reader reliabilities were good (RECIST1.1) to excellent (WHO and volumetric methods).

Conclusion

RECIST1.1, WHO and volumetric measures match with VCR after IC in patient with T3 laryngeal cancer. WHO criteria combine accuracy, reproducibility and practical use; they may be best adapted for response assessment and protocol design.

http://bit.ly/2UqZVGk