CHICAGO – A preoperative risk model accurately identifies women with endometrial cancer who are unlikely to have lymph node metastases, finds a prospective cohort study reported at the annual Meeting of the Society of Gynecologic Oncology.
The model – which uses favorable MRI features, endometrioid histology on biopsy, and a cancer antigen 125 (CA125) level of 35 U/mL or lower to define a low-risk group – had a negative predictive value of 97% when tested in 529 Asian women.
Susan London/Frontline Medical News
Dr. Sokbom Kang
“Using [our model], we can reliably identify patients with a low risk for lymph node metastasis before surgery,” said Dr. Sokbom Kang, director of the division of gynecologic oncology at the National Cancer Center, Goyang, Korea.
“In the clinic, our preoperative risk assessment may be useful in patient counseling. By sharing this risk information with our patients, we may improve their decision-making process about their surgery,” he added. “Not only does it help patient counseling and surgical planning, but it also may be useful in patient selection for future surgical trials.”
Some guidelines have stopped recommending routine lymphadenectomy in patients with endometrial cancer, according to Dr. Kang. “However, some experts still endorse this procedure, even in low-risk patients. Their argument is, low-risk patients cannot be accurately identified because preoperative tests are inaccurate.”
The risk model was developed by the Korean Gynecologic Oncology group (J. Clin. Oncol. 2012;30:1329-34) and has since been validated in smaller single-nationality cohorts.
In the new study, known as PALME (Preoperative Risk Assessment for Lymph Node Metastasis in Endometrial Cancer), it was tested among consecutive women from 25 hospitals in Korea, Japan, and China who had a histologic diagnosis of endometrial cancer. Those with squamous cell carcinoma or sarcoma histologies were excluded.
The women underwent MRI and CA125 testing in the 4 weeks before surgery. They had surgical staging with pelvic lymphadenectomy, and para-aortic lymphadenectomy was recommended. The median number of nodes removed was 23.
Results showed that the model classified 51% of the patients as having a low risk of lymph node metastases, reported Dr. Kang, who disclosed that he had no relevant conflicts of interest.
On the basis of surgical findings, the model had a negative predictive value of 97%, corresponding to a false-negative predictive rate of just 3%, which was in line with earlier results seen in the smaller validation studies.
In a receiver operating characteristic curve analysis, the model had a summarized sensitivity of 91% and a summarized specificity of 54%.
The performance was similar when tumor grade was substituted for CA125 level, except that specificity decreased significantly. “The lower specificity means fewer patients will benefit from our selective lymphadenectomy strategy, so it impairs the cost-effectiveness of our strategy,” Dr. Kang said.
The model performed similarly as well as a model using features of the primary tumor drawn from the final pathology report. “This proves our preoperative risk assessment has similar accuracy to the postoperative risk assessment for identifying a low risk of lymph node metastasis,” he concluded.