NF-kB and COX-2 Expression in Endometrial Lesions and Cancer
The function of COX-2 in carcinogenesis has been examined in colorectal cancer and as a prognostic marker for ovarian, breast, and gastric cancer. Yet, in several cancers, including breast, prostate, skin, lung, and pancreas, NF-κB expression is linked to poorer clinical outcome. However, the immunohistochemical expression of NF-κB has not been examined in EPs, EH, and EIN.
Scoring systems for tumor markers are usually based on the proportion of positive tumor cells and staining intensity. However, the interpretation of staining intensity is highly subjective and can be affected by storage time and variations in protocols and fixation procedures. Computer-assisted automated analysis programs help eliminate the inherent variability in pathologist-based scores and can increase the sensitivity of measurements of protein expression. Thus, we chose to use an automated system in our study.
Type 1 EC (endometrioid histology) constitutes the majority of EC cases and is associated primarily with estrogen stimulation. The classic steroid signaling pathway entails the activation of estrogen-responsive elements in the genome by estrogen receptor.
A secondary mechanism steroid signaling involves protein-to-protein interactions, wherein a receptor and ligand activate transcription factors, such as NF-κB, activator protein 1, and specificity protein 1, which subsequently stimulate transcription. In addition, estrogen correlates with inflammatory environments, stimulating the expression of inflammatory cytokines, such as interleukin (IL) 1, tumor necrosis factor α, matrix metallopeptidases, and IL-1b. A direct interaction between NF-κB and COX-2 transcription has also been proposed—the COX-2 promoter contains a binding site for NF-κB.
Mizumoto et al recently demonstrated that, despite its central role in endometrial carcinogenesis, the conventional KRAS-ERK1/2 pathway is insufficient to effect endometrial carcinogenesis and that NF-κB is a critical target of KRAS-induced endometrial carcinogenesis.
We noted significant differences in mean COX-2 expression in various nonmalignant lesions (EPs, EH) and EC. Mean COX-2 expression decreased progressively from EH to EIN and EC. In our series, we observed higher expression of COX-2 in EH than in EPs. Although polyps are not premalignant lesions, COX-2 expression might be attributed to the intense inflammatory process during its pathogenesis. This hypothesis has been corroborated by Cicinelli et al, who reported in 96 cases that COX-2 expression rose in 93.7% of small polyps with signs of chronic endometritis. Interestingly, COX-2 expression may differ according to the menopausal status, as reported by Erdemoglu et al, who found significantly higher expression in premenopausal polyps compared with postmenopausal polyps.
Instead of analyzing the normal endometrium, we opted to include benign polyps because polyps are benign tumors but not premalignant. Only a few authors have analyzed COX-2 expression in both nonneoplastic endometrium and cancer. Erkanli et al evaluated 50 ECs, 30 simple hyperplasias, and 20 proliferative endometria and found higher expression of COX-2 in EC and EH compared with proliferative endometrium. Similarly, Cao et al suggested that COX-2 was not expressed in the nonneoplastic endometrium, minimally expressed in the grade 1 endometrioid carcinomas, and stained most strongly in the grade 2 and grade 3 endometrioid carcinomas. Conversely, Orejuela et al did not find a statistically significant difference in COX-2 expression among EC (n = 14), all types of EH (n = 19), and normal endometrium (n = 10). However, the small number of patients may have impaired the results of that study.
We did not observe any correlation between COX-2 expression in EC and depth of invasion and histologic grade, for which conflicting data exist. Ferrandina et al reported that well-differentiated tumors with superficial myometrial invasion had lower expression of COX-2 compared with high-grade tumors or deep myometrial invasion. Fowler et al and Lambropoulou et al also found that COX-2 expression correlated with higher histologic grade. The last three studies included nonendometrioid histology in their analysis. Similarly to our study, Erkanli et al and Jeon et al did not find any correlation between COX-2 expression and depth of invasion or histologic grade.
The prognostic value of COX-2 expression in EC is controversial. We did not find COX-2 expression to be a prognostic factor, in contrast with Lambropoulou et al, who suggested that COX-2 expression was a prognostic factor but only in univariate analysis. Our findings corroborate with most studies, in which COX-2 expression had no effect on prognosis.
Regarding the prognostic value of NF-κB p50 and NF-κB p65, we also did not find any correlation with progression-free and overall survival. As far as we know, our study is the first to analyze NF-κB expression as a prognostic factor in EC. Interestingly, we also found that grade 3 tumors had a higher mean expression of NF-κB p65, and higher body mass index was associated with higher expression of NF-κB p50.
The decreased immunohistochemical expression of COX-2 in EC vs nonmalignant lesions might be explained by the presence of microsatellite instability (MSI). COX-2 expression can be suppressed when there is a defect in mismatch repair genes and MSI. Moreover, MSI appears to occur late in EC carcinogenesis—it is usually present in type 1 EC and is infrequent in EPs and EH without atypia. Further studies should be performed to confirm this hypothesis because we did not evaluate MSI in our series.
Our data also demonstrate greater expression of NF-κB p50 and NF-κB p65 in nonmalignant endometrial lesions compared with EC. We noted a significant and positive correlation between mean expression of NF-κB and COX-2 in EC, indicating that when mean COX-2 expression decreases, NF-κB also declines.
During endometrial carcinogenesis, endometrial tissue is exposed to various stimuli, and proliferative changes occur over time. Certain nonmalignant and premalignant lesions have more intense inflammatory microenvironments, some of which can undergo progressive alterations in histologic complexity and atypia, which culminate in EC.
The oncogenic and signal transduction pathways in endometrial carcinogenesis remain unknown, and the reason why COX-2 and NF-κB expression decline in EC compared with nonmalignant lesions is undetermined. However, COX-2 and NF-κB are clearly expressed in premalignant lesions (hyperplasia) and EC; thus, this signaling pathway merits further examination in endometrial carcinogenesis. New information regarding the molecular mechanisms of endometrial carcinogenesis can guide the future development of therapeutics for cancer by suppressing the carcinogenic pathway.
The heterogeneity of the studies and their methods make any comparison difficult, and it is important to note that we used TMA instead of whole tissue section for immunohistochemical staining. Among other studies, only Fowler et al also used TMA; however, they used a 0.6-mm tissue core while we used a 1-mm tissue core. Although it is argued that a single core sample per tumor may not be representative of the whole tumor, results using even one sample tumor have suggested an association between molecular features and clinicopathologic variables.
In contrast to other studies, we performed a novel technique of COX-2 and NF-κB immunohistochemical analysis using an automated system. A potential advantage of this method is that it provides a quantitative measure that distinguishes slight differences in staining intensity and helps to eliminate inherent variability in pathologist-based scores.
In conclusion, cytoplasmic immunohistochemical expression of COX-2 and NF-κB is lower in EC than in nonmalignant endometrial lesions, and COX-2 and NF-κB have no prognostic value in EC. Our data are the first findings to compare the immunohistochemical expression of NF-κB in EC and its precursor lesions and to study its prognostic value in EC.
Discussion
The function of COX-2 in carcinogenesis has been examined in colorectal cancer and as a prognostic marker for ovarian, breast, and gastric cancer. Yet, in several cancers, including breast, prostate, skin, lung, and pancreas, NF-κB expression is linked to poorer clinical outcome. However, the immunohistochemical expression of NF-κB has not been examined in EPs, EH, and EIN.
Scoring systems for tumor markers are usually based on the proportion of positive tumor cells and staining intensity. However, the interpretation of staining intensity is highly subjective and can be affected by storage time and variations in protocols and fixation procedures. Computer-assisted automated analysis programs help eliminate the inherent variability in pathologist-based scores and can increase the sensitivity of measurements of protein expression. Thus, we chose to use an automated system in our study.
Type 1 EC (endometrioid histology) constitutes the majority of EC cases and is associated primarily with estrogen stimulation. The classic steroid signaling pathway entails the activation of estrogen-responsive elements in the genome by estrogen receptor.
A secondary mechanism steroid signaling involves protein-to-protein interactions, wherein a receptor and ligand activate transcription factors, such as NF-κB, activator protein 1, and specificity protein 1, which subsequently stimulate transcription. In addition, estrogen correlates with inflammatory environments, stimulating the expression of inflammatory cytokines, such as interleukin (IL) 1, tumor necrosis factor α, matrix metallopeptidases, and IL-1b. A direct interaction between NF-κB and COX-2 transcription has also been proposed—the COX-2 promoter contains a binding site for NF-κB.
Mizumoto et al recently demonstrated that, despite its central role in endometrial carcinogenesis, the conventional KRAS-ERK1/2 pathway is insufficient to effect endometrial carcinogenesis and that NF-κB is a critical target of KRAS-induced endometrial carcinogenesis.
We noted significant differences in mean COX-2 expression in various nonmalignant lesions (EPs, EH) and EC. Mean COX-2 expression decreased progressively from EH to EIN and EC. In our series, we observed higher expression of COX-2 in EH than in EPs. Although polyps are not premalignant lesions, COX-2 expression might be attributed to the intense inflammatory process during its pathogenesis. This hypothesis has been corroborated by Cicinelli et al, who reported in 96 cases that COX-2 expression rose in 93.7% of small polyps with signs of chronic endometritis. Interestingly, COX-2 expression may differ according to the menopausal status, as reported by Erdemoglu et al, who found significantly higher expression in premenopausal polyps compared with postmenopausal polyps.
Instead of analyzing the normal endometrium, we opted to include benign polyps because polyps are benign tumors but not premalignant. Only a few authors have analyzed COX-2 expression in both nonneoplastic endometrium and cancer. Erkanli et al evaluated 50 ECs, 30 simple hyperplasias, and 20 proliferative endometria and found higher expression of COX-2 in EC and EH compared with proliferative endometrium. Similarly, Cao et al suggested that COX-2 was not expressed in the nonneoplastic endometrium, minimally expressed in the grade 1 endometrioid carcinomas, and stained most strongly in the grade 2 and grade 3 endometrioid carcinomas. Conversely, Orejuela et al did not find a statistically significant difference in COX-2 expression among EC (n = 14), all types of EH (n = 19), and normal endometrium (n = 10). However, the small number of patients may have impaired the results of that study.
We did not observe any correlation between COX-2 expression in EC and depth of invasion and histologic grade, for which conflicting data exist. Ferrandina et al reported that well-differentiated tumors with superficial myometrial invasion had lower expression of COX-2 compared with high-grade tumors or deep myometrial invasion. Fowler et al and Lambropoulou et al also found that COX-2 expression correlated with higher histologic grade. The last three studies included nonendometrioid histology in their analysis. Similarly to our study, Erkanli et al and Jeon et al did not find any correlation between COX-2 expression and depth of invasion or histologic grade.
The prognostic value of COX-2 expression in EC is controversial. We did not find COX-2 expression to be a prognostic factor, in contrast with Lambropoulou et al, who suggested that COX-2 expression was a prognostic factor but only in univariate analysis. Our findings corroborate with most studies, in which COX-2 expression had no effect on prognosis.
Regarding the prognostic value of NF-κB p50 and NF-κB p65, we also did not find any correlation with progression-free and overall survival. As far as we know, our study is the first to analyze NF-κB expression as a prognostic factor in EC. Interestingly, we also found that grade 3 tumors had a higher mean expression of NF-κB p65, and higher body mass index was associated with higher expression of NF-κB p50.
The decreased immunohistochemical expression of COX-2 in EC vs nonmalignant lesions might be explained by the presence of microsatellite instability (MSI). COX-2 expression can be suppressed when there is a defect in mismatch repair genes and MSI. Moreover, MSI appears to occur late in EC carcinogenesis—it is usually present in type 1 EC and is infrequent in EPs and EH without atypia. Further studies should be performed to confirm this hypothesis because we did not evaluate MSI in our series.
Our data also demonstrate greater expression of NF-κB p50 and NF-κB p65 in nonmalignant endometrial lesions compared with EC. We noted a significant and positive correlation between mean expression of NF-κB and COX-2 in EC, indicating that when mean COX-2 expression decreases, NF-κB also declines.
During endometrial carcinogenesis, endometrial tissue is exposed to various stimuli, and proliferative changes occur over time. Certain nonmalignant and premalignant lesions have more intense inflammatory microenvironments, some of which can undergo progressive alterations in histologic complexity and atypia, which culminate in EC.
The oncogenic and signal transduction pathways in endometrial carcinogenesis remain unknown, and the reason why COX-2 and NF-κB expression decline in EC compared with nonmalignant lesions is undetermined. However, COX-2 and NF-κB are clearly expressed in premalignant lesions (hyperplasia) and EC; thus, this signaling pathway merits further examination in endometrial carcinogenesis. New information regarding the molecular mechanisms of endometrial carcinogenesis can guide the future development of therapeutics for cancer by suppressing the carcinogenic pathway.
The heterogeneity of the studies and their methods make any comparison difficult, and it is important to note that we used TMA instead of whole tissue section for immunohistochemical staining. Among other studies, only Fowler et al also used TMA; however, they used a 0.6-mm tissue core while we used a 1-mm tissue core. Although it is argued that a single core sample per tumor may not be representative of the whole tumor, results using even one sample tumor have suggested an association between molecular features and clinicopathologic variables.
In contrast to other studies, we performed a novel technique of COX-2 and NF-κB immunohistochemical analysis using an automated system. A potential advantage of this method is that it provides a quantitative measure that distinguishes slight differences in staining intensity and helps to eliminate inherent variability in pathologist-based scores.
In conclusion, cytoplasmic immunohistochemical expression of COX-2 and NF-κB is lower in EC than in nonmalignant endometrial lesions, and COX-2 and NF-κB have no prognostic value in EC. Our data are the first findings to compare the immunohistochemical expression of NF-κB in EC and its precursor lesions and to study its prognostic value in EC.
SHARE
