Clinical relevance of sonometry-derived tumour thickness in malignant melanoma--a statistical analysis

Abstract

High-frequency sonography has been shown to be a useful tool in planning operative strategy in the surgery of malignant melanoma (MM). The purpose of the present study was to compare sonometric and histometric data of tumour thickness in primary cutaneous MM, applying statistical methods in order to evaluate the pre-operative relevance of sonometry. The thickness of 259 melanomas was measured preoperatively by a 20-MHz B scan, and postoperatively by histometry. Statistical analysis was performed using Pearson's correlation coefficient and absolute and relative differences. Although the correlation between sonometry and histometry was good (r = 0.88), there was a mean difference of 0.39 mm (relative difference 28%). Overall, sonometry was in agreement with the corresponding histological classes in 75% of cases. However, tumours assessed by ultrasound as between 0.55 and 0.95 mm thick were incorrectly classified according to histology in 34%, and those between 1.30 and 1.70 mm were incorrectly classified in 50% of cases. Our data reveal greater differences between sonometry and histometry using appropriate statistical methods. A concept to assess differences between sonometry and histometry is recommended.