Clinical application of body diffusion weighted MR imaging in the diagnosis and preoperative N staging of cervical cancer.

OBJECTIVE

To evaluate the clinical impact of body diffusion weighted imaging (DWI) on the diagnosis and preoperative N staging of cervical cancer.

METHODS

Twenty-four patients (mean age 37.9 years old) with proved cervical cancer by cervical biopsy and 24 female patients with other suspected pelvic abnormalities received preoperative body DWI scan. Results of body DWI were compared with pathological findings. The apparent diffusion coefficient (ADC) values of normal cervix and different pathological types of cervical cancer were compared. ADC value of normal or inflammatory lymph nodes was also compared with that of metastatic ones. Student's t test was used for statistical analysis.

RESULTS

There were 5 adenocarcinomas and 19 epitheliomas showed with biopsy results, and DWI showed 21 cervical lesions out of them (87.5%). ADC values of the normal cervix (n = 24), epithelioma (n = 19), and adenocarcinoma (n = 5) were (1.73 +/- 0.31) x 10(-3), (0.88 +/- 0.22) x 10(-3), and (1.08 +/- 0.12) x 10(-3) mm2/s, respectively. Statistical analysis showed significant difference in ADC value between normal cervical tissue and either tumor tissues (both P < 0.01). In patients had lymphadenectomy (n = 24), totally 67 lymph nodes including 16 metastatic lymph nodes were pathologically analyzed, and DWI showed 66 (98.5%) out of them. ADC values of normal/inflammatory and metastatic lymph nodes were (1.07 +/- 0.16) x 10(-3) and (0.77 +/- 0.13) x 10(-3) mm2/s (P < 0.01). Receiver operating characteristic (ROC) curve of ADC value of metastatic lymph node showed that area under curve was 0.961.

CONCLUSIONS

ADC value in cervical carcinoma is lower than that in normal cervix, and ADC may have predictive value in subtype discrimination. ADC value may improve the preoperative characterization of lymph node metastasis. And at least abdominal and pelvic DWI scan is suggested for N staging evaluation in such patients.