OBJECTIVES: To assess the utility of tumor diameter (TD) for predicting lymphatic dissemination (LD) and determining need for lymphadenectomy following diagnosis of endometrioid endometrial cancer.

METHODS: Patients diagnosed with stage I-III endometrioid endometrial cancer during 2003-2013 who underwent complete lymphadenectomy during hysterectomy were studied. Intraoperative predictors of LD included TD, grade, myometrial invasion (MI), age, body mass index, and race/ethnicity. Candidate logistic regression models of LD were evaluated for model fit and predictive power.

RESULTS: Of 737 cancer patients, 68 (9.2%) were node-positive. Single-variable models with only continuous TD (c-statistic 0.77, 95% CI 0.71-0.83) and dichotomous TD with 50-mm cut-off (TD50; c-statistic 0.73, 95% CI 0.67-0.78) were significantly more predictive than with the standard 20-mm cut-off (c-statistic 0.56, 95% CI 0.53-0.59). Overall, the most important LD predictors were TD50 and MI3rds (three-category form). The best candidate model (c-statistic 0.84, 95% CI 0.80-0.88) suggested odds of LD were five times greater for TD >50mm than ≤50mm (OR 4.91, 95% CI 2.73-8.82) and six and ten times greater for MI >33% to ≤66% (OR, 5.70; 95% CI, 2.25-14.5) and >66% (OR 10.2, 95% CI 4.11-25.4), respectively, than ≤33%. Best-model false-negative (0%) and positive (57.2%) rates demonstrated marked improvement over traditional risk-stratification false-negative (1.5%) and positive (76.2%) rates (c-statistic 0.77, 95% CI 0.72-0.82).

CONCLUSIONS: Tumor diameter is an important predictor of LD. Our risk model, containing modified forms of MI and TD, yielded a lower false-negative rate and can significantly decrease the number of lymphadenectomies performed on low-risk women.