We appreciate Dr. Troost et al.'s interest and thoughtful comments on our article (1). Our study was intended to investigate the relationship between the metabolic tumor volumes (MTV) measured on 18F-fluorodeoxyglucose positron emission tomography (18FDG-PET) and short-term outcome in patients with pharyngeal cancers treated by nonsurgical modalities. We would like to respond to each of his points.
Dr. Troost et al. noted that a threshold of 2.5 for the standardized uptake value (SUV) for segmentation of the 18FDG-PET is arbitrary and protocol dependent.
Basically, we totally agree that the most optimal threshold for the MTV in 18FDG-PET segmentation in head and neck tumors remains to be determined.
Our study was meant to present clinical data about 18FDG-PET–based MTVs in an effort to find the ideal method for head and neck cancers. To minimize the methodologic errors, we took several, cautious steps before coming up with a conclusion.
First, we tested different measurement thresholds including isocontours of SUV 3.0 and 3.5 and fixed percentage (40%) of the maximum SUV (data not presented in the report). However, we identified a major discordance in lesion delineation between 18FDG-PET images and corresponding contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI). Furthermore, the cutoff for the MTVs from SUV 2.5 was not deducted arbitrarily, but from concrete statistical analyses.
Second, all measurements were done by determining the region of interest (ROI) directly on the CT-attenuated PET images, which incorporate the enhanced anatomic information (2). To reduce the confounding influence by surrounding, physiologically glucose-avid tissues in head and neck, coronal, axial, and sagittal planes were reviewed at the same time to adjust the boundary of ROI. Unlike Troost et al.'s study, there was no case of measurement failure with an isocontouring of SUV 2.5 in our study (3).
Third, volumetric measurement with the automatic ROI method used in our study has several advantages, for instance, less chance of interobserver/intraobserver variability and more applicability in clinical practice.
Dr. Troost et al. mentioned that a threshold of SUV 2.5 is ineffective for the segmentation of metastatic lymph nodes. We admit that he exactly pointed out the problem we have wrestled with. However, we also assumed that incorporating all the metastatic lesions in the body is reasonable in assessing the role of the MTVs as predictor for treatment outcome, especially in patients, most of whom (90.3%) had undergone systemic as well as local therapy (concurrent chemoradiation). We used not only 18FDG-PET findings but also information from physical examination, CT, or MRI to determine the target lesions to reduce the possible, false-positive measurement of the MTVs of lymph nodes.
Dr. Troost et al. commented that patients with necrotic areas within the tumor or lymph nodes must be excluded because of their confounding influence on the correlation of MTVs with short-term outcome.
Among the patients enrolled in our study, five patients (four oropharynx and one hypopharynx) showed FDG uptake-void areas within lymph nodes, as well as in CT or MRI. However, it is not feasible to distinguish the cystic degeneration of metastatic lymph nodes, which is not an infrequent finding in oropharyngeal cancer, from true necrotic change from tumor hypoxia in imaging study (4). Moreover, it is less likely that a small fraction of these patients might have affected the interpretation of our analyses.
Dr. Troost et al. stated that analyses of correlation between gross tumor volume on CT and the MTVs and short-term treatment outcomes must have been included in the study.
We agree with his comment that it is also another limitation of our study and a good source of future research. Although the direct values measured on CT were not obtained in our study, we used clinical T stage, which is based on physical examination and image findings, to present the correlation analyses between TNM stage and the MTVs. According to our data, a positive correlation was identified between these two values.
Recently, the MTVs are attracting a great deal of attention because of increased diagnostic accuracy of 18FDG-PET/CT and the availability of more advanced measurement methods. We agree with Dr. Troost et al. that iterative validation and cautious application of methods are required in measuring the MTVs for head and neck tumors and regional lymph nodes. Also, we want to point out that feasibility is the key element to introduce the MTVs into clinical practice, not affected by institutional-based specific protocol.
RSS Feeds