ABSTRACT
To examine correlations among nuclear, architectural, and International Federation of Gynecology and Obstetrics (FIGO) grading systems, and their relationships with lymph node (LN) involvement in endometrioid endometrial cancer.
Histopathology slides of 135 consecutive patients were reviewed with respect to tumor grade and LN metastasis. Notable nuclear atypia was defined as grade 3 nuclei. FIGO grade was established by raising the architectural grade (AG) by one grade when the tumor was composed of cells with nuclear grade (NG) 3. Correlations between the grading systems were analyzed using Spearman’s rank correlation coefficients, and relationships of grading systems with LN involvement were assessed using logistic regression analysis.
Correlation analysis revealed a significant and strongly positive relationship between FIGO and architectural grading systems (r=0.885, p=0.001); however, correlations of nuclear grading with the architectural (r=0.535, p=0.165) and FIGO grading systems (r=0.589, p=0.082) were moderate and statistically non-significant. Twenty-five (18.5%) patients had LN metastasis. LN involvement rates differed significantly between tumors with AG 1 and those with AG 2, and tumors with FIGO grade 1 and those with FIGO grade 2. In contrast, although the difference in LN involvement rates failed to reach statistical significance between tumors with NG 1 and those with NG 2, it was significant between NG 2 and NG 3 (p=0.042). Although all three grading systems were associated with LN involvement in univariate analyses, an independent relationship could not be established after adjustment for other confounders in multivariate analysis.
Nuclear grading is significantly correlated with neither architectural nor FIGO grading systems. The differences in LN involvement rates in the nuclear grading system reach significance only in the setting of tumor cells with NG 3; however, none of the grading systems was an independent predictor of LN involvement.
Keywords: Endometrial cancer, grade, lymph node involvement
Introduction
Material and Methods
Results
A total of 135 patients were enrolled in the analysis. The majority of patients had AG 1 (56.3%), NG 2 (45.9%), and FIGO grade 1 (54.1%) tumors. Eighty (59.3%) patients had pelvic lymphadenectomy alone, and 55 (40.7%) had combined pelvic and paraaortic lymphadenectomy. LN involvement was identified in 25 (18.5%) patients (Table 1).
Correlation analysis revealed that there was a significant and very strongly positive relationship between the FIGO and architectural grading systems (r=0.885, p=0.001); however, correlations of nuclear grading with the architectural (r=0.535, p=0.165) and FIGO grading systems (r=0.589, p=0.082) were moderate and statistically non-significant (Table 2).
The rates of LN involvement according to each grading system are summarized in Table 3. LN involvement was detected in 7.9% of tumors with AG 1, 25.0% of tumors with AG 2, and 47.3% of tumors with AG 3. LN involvement rates according to FIGO grades were as follows: 5.4% for grade 1, 31.6% for grade 2, and 37.5% for grade 3. LN involvement rates differed significantly between tumors with AG 1 and AG 2 (p=0.045), and between tumors with FIGO grades 1 and 2 (p=0.031), whereas there were no significant differences between AG 2 and AG 3 (p=0.069), and between FIGO grades 2 and 3 (p=0.327).
The rates of LN involvement based on nuclear grading system were as follows: 6.2% for NG 1, 20.9% for NG 2, and 36.0% for NG 3. In contrast to architectural and FIGO grading systems, the difference in LN involvement rates failed to reach statistical significance between tumors with NG 1 and those with NG 2 (p=0.115), but it was significant between NG 2 and NG 3 (p=0.042) (Table 3).
In order to assess independent relationships between grading systems and LN metastasis, two different logistic regression models were developed because a strong correlation between FIGO grade and AG would confound the possible associations (Table 4). NG and AG were assigned to the first model, and the FIGO grade was separately evaluated in the second model. Both models also included deep myometrial invasion and lymphovascular space involvement (LVSI) as potential covariates. Although all three grading systems were associated with LN involvement in univariate analyses, an independent relationship could not be established after adjustment for other confounders in multivariate analyses. LVSI was consistently the sole independent predictor of LN metastasis in multivariate analyses (p=0.001).
Discussion
The lack of an objective definition for “notable nuclear atypia” and the moderate inter-observer agreement in distinction of squamous from non-squamous solid growth in the FIGO grading system led to the proposal of alternative binary grading systems by some researchers over the past two decades (7,8,9). Lax et al. (7) described a binary grading system that uses a low magnification evaluation of the presence of necrosis, pattern of invasion, and amount of solid growth to divide endometrioid ECs into low- and high-grade tumors. The authors suggested that a tumor should be considered as high-grade when it exhibits at least two of the following features: i) more than 50% solid growth (without distinction of squamous from non-squamous epithelium); ii) a diffusely infiltrative, rather than expansive, growth pattern; and iii) tumor cell necrosis. The authors reported that both inter- and intraobserver agreements using the binary grading system were superior compared with the FIGO and nuclear grading systems. Scholten et al. (8) conducted a study to compare the reproducibility of FIGO grading system with the novel binary grading system proposed by Lax et al. (7); however, they found that the inter-observer agreement for both systems was moderate, with 70% and 73% agreement rates for the FIGO and binary grading systems, respectively. The authors proposed that if a simple architectural binary grading system that divides tumors into low- and high-grade based solely on the proportion of solid tumor growth (≤50% or ˃50%) was used in the grading of ECs, a much better agreement rate (85%) could be achieved. In another alternative binary grading system (low-grade vs. high-grade), Alkushi et al. (9) suggested that tumors should be considered high-grade in the presence of at least two of the following criteria: i) predominantly papillary or solid growth pattern, ii) mitotic index ≥6/10 high power fields, and iii) severe nuclear atypia. The authors reported that this system had more prognostic power than the three-tiered FIGO and binary system of Lax et al. (7) when applied to all tumors regardless of tumor histotype; however, the FIGO grading system was superior for prognostication when only endometrioid type ECs were considered.
Currently, none of these alternative systems has become widespread because it is not clear whether they would significantly improve the prognostic utility of the current method (10). Moreover, in a recent study comparing new binary systems with the existing three-tiered FIGO grading system, Guan et al. (11) demonstrated that the FIGO grading system using the nuclear criteria of Zaino et al. (4) was prognostically superior to the other systems, particularly in patients with endometrioid-type EC.
