Detection of lymph node metastases in head and neck cancer: a meta-analysis comparing US, USgFNAC, CT and MR imaging.

PURPOSE To perform a meta-analysis comparing ultrasonography (US), US guided fine needle aspiration cytology (USgFNAC), computed tomography (CT), and magnetic resonance imaging (MRI) in the detection of lymph node metastases in head and neck cancer. METHODS MEDLINE, EMBASE and Cochrane databases were searched (January 1990-January 2006) for studies reporting diagnostic performances of US, USgFNAC, CT, and MRI to detect cervical lymph node metastases. Two reviewers screened text and reference lists of potentially eligible articles. Criteria for study inclusion: (1) histopathology was the reference standard, (2) primary tumors and metastases were squamous cell carcinoma and (3) data were available to construct 2 x 2 contingency tables. Meta-analysis of pairs of sensitivity and specificity was performed using bivariate analysis. Summary estimates for diagnostic performance used were sensitivity, specificity, diagnostic odds ratios (DOR) (95% confidence intervals) and summary receiver operating characteristics (SROC) curves. RESULTS From seventeen articles, 25 data sets could be retrieved. Eleven articles studied one modality: US (n=4); USgFNAC (n=1); CT (n=3); MRI (n=3). Six articles studied two or more modalities: US and CT (n=2); USgFNAC and CT (n=1); CT and MRI (n=1); MRI and MRI-USPIO (Sinerem) (n=2); US, USgFNAC, CT and MRI (n=1). USgFNAC (AUC=0.98) and US (AUC=0.95) showed the highest areas under the curve (AUC). MRI-USPIO (AUC=0.89) and CT (AUC=0.88) had similar results. MRI showed an AUC=0.79. USgFNAC showed the highest DOR (DOR=260) compared to US (DOR=40), MRI-USPIO (DOR=21), CT (DOR=14) and MRI (DOR=7). Conclusion USgFNAC showed to be the most accurate imaging modality to detect cervical lymph node metastases.