Comparative study of three sonoelastographic scores for differentiation between benign and malignant cervical lymph nodes.

PURPOSE

The aim of the study was to explore the diagnostic value of three different sonoelastographic scoring systems (labeled S1-S3) for the differentiation between benign and malignant cervical lymph nodes.

MATERIALS AND METHOD

The authors propose a six pattern scoring system of the elastographic images with pattern 1 - representing purely soft nodes, pattern 2 - predominantly soft nodes, pattern 3 - predominantly soft nodes with focal had area, pattern 4 - predominantly hard node, pattern 5 - entirely hard node and pattern 6 - node with necrosis. The sonoelastographic images of 50 benign and 70 malignant lymph nodes were assessed. The area under the ROC curve (AUROC) for the differentiation between benign vs. malignant and benign vs. metastatic nodes were analyzed for the three scoring systems.

RESULTS

When all the malignant lymph nodes were considered, the S1 score showed an AUROC=0.873 (95%CI [0.805-0.918], where CI=confidence interval; p<0.001), sensibility (Se)=58.57%, and specificity (Sp)=96%. For S2 score the AUROC was 0.890 (95%CI [0.824-0.933], p<0.001), Se=92.86%, and Sp=72%. For S3 score, the AUROC was 0.852 (95%CI [0.778-0.902], p<0.001), Se=64.29%, and Sp=94%). When lymphomatous nodes were excluded, for S1 the AUROC was 0.884 (95%CI [0.809-0.932], p<0.001), Se=64%, and Sp=96%. For S2 the AUROC was 0.894 (95%CI [0.818-0.939], p<0.001), Se=92%, and Sp=72%. For S3, the AUROC was 0.856 (95%CI [0.771-0.911], p<0.001), Se=66%, and Sp=94%. In the S3 scoring system, setting the benign vs. malignant cut off at pattern 3 increases the sensibility (41-65%) with minimal loss of specificity (96-94%). From the gray-scale and Doppler criteria, changes of the nodular margins and the presence of the vessels in the cortical part of the lymph node showed both very high sensibility and specificity, the others criteria taken into account had either very good sensibility with low specificity or high specificity and low sensibility.

CONCLUSIONS

Our study suggests that there are no significant differences between the three scoring systems in terms of overall diagnostic value.