Abstract
98
Objectives Accurate assessment of cervical lymph nodes (LNs) for determination of metastatic involvement is of prime importance for management. Our objective was to investigate the role of combinatorial evaluation of metabolic and morphologic imaging to enhance the ability to distinguish between benign and malignant LNs.
Methods Forty four adult patients with head and neck cancer (n=42) or lymphoma (n=2) were studied. PET/CT and neck ceCT obtained simultaneously or sequentially at initial (n=14) or restaging (n=33). Fifty two LNs were evaluated based on SUVmax, size, shape (elliptical vs round), extracapsular extension(ecext-irregular margins), necrosis and fatty hilum. Histopathology (n=40) and 12 month follow-up (n=4) were used for confirmation. The analyses included ROC curves and two-tailed Fischer exact probability test.
Results The optimal SUV cut-off to differentiate between benign and malignant LNs was 3.6 (p=0.0001 ) with a sensitivity and specificity of 88% and 70%, respectively. Among ceCT variables only shape, necrosis and ecext were significant (Table 1). The combined analyses showed a sensitivity and specificity of 91% & 100% for SUV/ecext, 100% & 33.3% for SUV/necrosis and 86.7% & 42.9% for SUV/shape. With a SUV cutoff of 5.2, the combined analyses showed sensitivities and specificities of 63.6% & 100% for SUV/ecext, 83% & 100% for SUV/necrosis, 80% & 85% for SUV/shape.
Conclusions The combination of SUVmax with the presence of ecext provides the best diagnostic information for identifying LN metastasis. Only when SUVmax cutoff is kept high, the presence of necrosis and irregular margins yield high sensitivity and specificity
Ecext: extracapsular spread
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.