Tateishi Tsutomu, Machi Junji, Feleppa Ernest J, Oishi Andrew J, Furumoto Nancy L, Oishi Robert H, McCarthy Laurence J, Yanagihara Eugene, Shirouzu Kazuo
Department of Surgery, University of Hawaii at Manoa, John A. Burns School of Medicine and Kuakini Medical Center, 405 North Kuakini Street, Suite 601, Honolulu, Hawaii 96817-6301, USA.
J Clin Ultrasound. 2004 Jan;32(1):1-7. doi: 10.1002/jcu.10210.
The aim of this in vitro study was to assess the feasibility of using high-frequency sonography to identify colorectal lymph nodes and to diagnose colorectal lymph node metastasis.
In part 1 of this study, resected colorectal tissues from 13 patients with colorectal cancer were scanned in a water bath using B-mode sonography performed at high frequency (10 MHz) to identify lymph nodes. The colorectal tissues were then carefully dissected to remove all lymph nodes. Detectability was calculated as the ratio of the number of sonographically detected nodes to the total number of histopathologically confirmed nodes. Student's t test was performed to compare sizes between these groups; a p value of less than 0.05 was considered significant. In part 2, 4 features of lymph nodes identified on B-mode sonography--size, shape, border, and echogenicity--and their combinations were evaluated for their ability to diagnose lymph node metastasis. Discriminant and receiver operating characteristic curve analyses were performed.
In part 1, B-mode sonography performed in vitro detected 79 (48%) of the 165 histopathologically identified lymph nodes and 34 (87%) of the 39 histopathologically identified metastatic nodes. The mean size, or mean longest axis (+/- standard deviation), of the sonographically detected nodes (6.4 +/- 2.9 mm) was significantly larger than that of undetected nodes (3.6 +/- 1.7 mm; p < 0.01). In part 2, the most effective feature distinguishing metastatic from nonmetastatic lymph nodes was echogenicity, followed by size, shape, and border. However, a combination of at least 2 features (eg, echogenicity and size) provided better distinction of nodes than did any 1 feature. In the receiver operating characteristic curve of the 4-feature combination, an increase in sensitivity is accompanied by a decrease in specificity: at a sensitivity of 100%, specificities decreased to 60% or less. However, even with the optimal combination of features, the sensitivity and specificity did not both reach 85% at any operating point.
The results of this node-by-node in vitro study show the current limitations and potential of sonography for assessing colorectal lymph nodes. High-frequency sonography may be insufficient for identifying lymph node metastasis in colorectal cancer.
