The reproducibility of histological parameters employed in the novel binary grading systems of endometrial cancer

To compare the interobserver reproducibility and prognostic value of the FIGO grading system with the histological parameters employed in the various recently proposed binary grading systems of endometrial cancer.

Seventy two consecutive stage I endometrioid endometrial carcinomas from hysterectomy specimens were independently graded by two pathologists. Clinical data and outcome were obtained from the patients' records. The following histological parameters were evaluated: FIGO grade (dichotomized to grades 1 and 2 vs. grade 3), nuclear atypia, presence of more than 50% solid growth, diffusely infiltrative rather than expansive growth pattern, presence of tumor cell necrosis, and mitotic count. Interobserver agreement was measured by the kappa (k) statistics. Kaplan-Meier survival analysis, log-rank tests and Cox proportional hazard regression were used to evaluate the equality of survival distributions and to model the overall effects of the various predictor variables on survival.

The interobserver reproducibility was as follows: FIGO grade, k=0.65; nuclear atypia, k=0.63; solid growth, k=0.51; infiltrative growth pattern, k=0.38; tumor necrosis, k=0.52; and mitotic index, k=0.44. In the comparison of the Kaplan-Meier curves, the following parameters were associated with a significantly poorer survival: FIGO grade 3, p=0.02; presence of more than 50% solid growth, p=0.01; and a high mitotic index, p=0.01. The other binary histological parameters were not significantly predictive of survival.

The proposed novel binary grading parameters are not advantageous in terms of interobserver reproducibility and prognostic significance over dichotomization to FIGO grades 1 and 2 vs. grade 3. A simple binary grade based solely on presence of more than 50% solid growth has a comparable reproducibility and prognostic value.