Tumor volume in pharyngolaryngeal squamous cell carcinoma: comparison at CT, MR imaging, and FDG PET and validation with surgical specimen.

Abstract

To compare computed tomography (CT), magnetic resonance (MR) imaging, and fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography (PET) for delineation of gross tumor volume (GTV) in pharyngolaryngeal squamous cell carcinoma and to validate results with the macroscopic surgical specimen when available. Twenty-nine patients with stages II-IV squamous cell carcinoma treated with radiation therapy or chemotherapy and radiation therapy (n = 20) or with total laryngectomy (n = 9) were enrolled. Ten patients had oropharyngeal, 13 had laryngeal, and six had hypopharyngeal tumors. CT, MR imaging, and PET were performed with patients immobilized in a customized thermoplastic mask, and images were coregistered. GTVs obtained with the three modalities were compared quantitatively and qualitatively. If patients underwent total laryngectomy, images were validated with the surgical specimen after three-dimensional coregistration. The effect of each modality was estimated with linear mixed-effects models. Adjustments for multiple comparisons were made with the Bonferonni or Sidak method. For oropharyngeal tumors and for laryngeal or hypopharyngeal tumors, no significant difference (P >.99) was observed between average GTVs delineated at CT (32.0 and 21.4 cm(3), respectively) or MR imaging (27.9 and 21.4 cm(3), respectively), whereas average GTVs at PET were smaller (20.3 [P </=.10] and 16.4 cm(3) [P </=.01], respectively). GTVs from surgical specimens were significantly smaller (12.6 cm(3), P </=.06). In nine patients for whom a surgical specimen was available, no modality adequately depicted superficial tumor extension; this was due to limitations in spatial resolution. In addition, false-positive results were seen for cartilage, extralaryngeal, and preepiglottic extensions. Compared with GTVs at CT and MR imaging, GTVs at FDG PET were smaller. In nine patients for whom a surgical specimen was available, PET was found to be the most accurate modality. However, no modality managed to depict superficial tumor extension.