Comparing risk stratification criteria for predicting lymphatic dissemination in endometrial cancer.

OBJECTIVE

To compare two published risk stratification models (Milwaukee Model vs. Mayo Criteria) to predict lymphatic dissemination (LD) in endometrioid endometrial cancer (EC).

METHODS

Patients with stage I-III EC undergoing surgery from 1/1/2004-9/30/2013 were retrospectively reviewed and classified as low-risk vs at-risk for LD using two independent risk models. LD was defined as positive nodes at surgery or lymph node recurrence within 2 years of surgery after negative lymph node dissection (LND) or when LND was not performed. False positive (FP) and false negative (FN) rates for each risk model were calculated.

RESULTS

Among 1103 patients, 81 (7.3%) had LD (72 positive LN and 9 LN recurrences), and most (90.2%) had stage I EC. The Milwaukee Model yielded a low at-risk rate for LD (38.1%) but a high FN rate (13.6%, 95% CI 7.0-23.0). The traditional Mayo Criteria using a cut-off of 2 cm for tumor diameter (TD) had a higher at-risk rate for LD (69.5%) but a FN rate of 0% (95% CI, 0-4.5). Modifying the Mayo Criteria using a TD cutoff of ≤3 cm identified fewer women at-risk (56.8% vs. 69.5%) and had a lower FP rate (53.6% vs. 67.1%), but had a higher FN rate (3.7%, 95% CI, 0.8-10.4).

CONCLUSIONS

The Milwaukee Model had the lowest at-risk rate of LD but an unacceptable FN rate. Modifying the Mayo Criteria by increasing the TD cutoff from the traditional ≤2 cm to ≤3 cm would spare an estimated 13.5% of patients LND, but the accompanying FN rate is unacceptably high. The traditional Mayo Criteria for low-risk EC remains the most sensitive in determining which patients LND can be omitted.