Impact of small node size on clinical interpretation of F18 FDG PET CT for primary head and neck SCC malignancy. Bad things can come in small packages.

Objectives: Evaluation of cervical nodal metastases is vital for pre-treatment staging of patients with head and neck squamous cell carcinoma (HNSCC). Integrated PET / CT has been well validated for evaluation of primary, nodal and distant disease in HNSCC patients. However, assessment of small volume nodes remains a dilemma. Few studies have evaluated the significance of quantitative assessment with SUV max for diagnosis of nodal metastases within the sub-centimetre nodal group. The aim of this study was to assess diagnostic accuracy of FDG-PET / CT in evaluation of cervical nodal metastases based on nodal size and by HPV status.

Methods: 38 patients with primary head and neck SCC were prospectively recruited over a 2 year period from 2014 to 2015. All patients underwent resection of the primary tumour and cervical node dissection following FDG-PET / CT evaluation at our institution. Surgical nodal dissection was in blocks according to node station. Neck dissection was unilateral in 8/38, and bilateral in 30/38 patients. F18 FDG-PET / CT images were acquired 60 minutes after injection of F18 FDG. The PET and CT images were read clinically by experienced readers blinded to the surgical results, and assessment of SUV max was undertaken in all visible nodes, to derive ROC curves for varying nodal sizes.

Results: 38 patients had 67 neck sides dissected (32 bilateral, 6 bilateral), involving 353 nodal levels and a total of 2700 lymph nodes (mean 40.3 lymph nodes per neck side). Histopathological analysis revealed lymph node metastases in 116 of the 2701 dissected lymph nodes (4.3%). Mean size of pathologically involved nodes was 11mm, of whom 64% were sub-centimeter. On nodal based analysis, overall clinical sensitivity, specificity and diagnostic odds ratio were 51.7%, 98.7% and 82.8% for PET / CT. Using cervical node levels based analysis, the sensitivity, specificity was 75% and 94%. Clinical sensitivity, specificity and diagnostic odds ratio of PET / CT for sub-centimeter nodes were 30.7%, 99.2% and 57.6 compared to 90.2%, 84.8% and 51.5 for supra-centimeter nodes. All nodes with SUV max >5.0 were all confirmed to be metastatic on histopathology. Using ROC analysis for sub-centimeter nodes, the optimal SUV max cutoff was determined at 2.3, yielding a sensitivity and specificity of 73.7% and 94.4%. By contrast, in 蠅1cm nodes, a higher SUV max, 4.9 achieved similar sensitivity and specificity of 73.2% and 99.1% respectively.

Conclusion: Sub-centimeter lymph nodes are the most prevalent involved node size in HNSCC. SUV max cutoffs of 2.3 demonstrated a superior sensitivity to clinical reading by experts, suggesting that we need to recalibrate interpretation of quantitative FDG findings in small nodes in patients with known malignancy, to improve diagnostic accuracy. Research Support: