An external validation study of nomograms designed to predict isolated loco-regional and distant endometrial cancer recurrences: how applicable are they?

BACKGROUND

To externally validate and assess the robustness of two nomograms to predict the recurrence risk of women with endometrial cancer (EC).

METHODS

Using an independent, multicentre external patient cohort we assessed the discrimination and calibration of two nomograms--the 3-year isolated loco-regional (ILRR) and distant (DR) recurrence nomograms--in women with surgically treated stage I-III EC.

RESULTS

Two hundred and seventy one eligible women were identified from two university hospital databases and the Senti-Endo trial. The median follow-up and initial recurrence time were 38.1 (range: 12-69) and 22.0 (range: 8.3-55) months, respectively. The overall recurrence rate was 13.8% (37 out of 271). Predictive accuracy according to the discrimination was 0.69 (95% CI, 0.58-0.79) and 0.66 (95% CI, 0.60-0.71) for the 3-year ILRR and DR nomograms, respectively. The correspondence between observed recurrence rate and the nomogram predictions suggests a moderate calibration of the nomograms in the validation cohort.

CONCLUSION

The nomograms were externally validated and shown to be partly generalisable to a new and independent patient population. The tools need to be improved by including information on the lymph node status and adjuvant therapies.