Plasma High Mobility Group Box 1 (HMGB1), Osteopontin (OPN), and Hyaluronic Acid (HA) as Admissible Biomarkers for Endometriosis

Abstract

Identification of biomarkers for endometriosis is an unmet medical need that demands to be fulfilled. In this study, we first used a mouse model of endometriosis and evaluated the potential utility of select biomarkers based on serial observations. Since fibrosis is the end result of lesional development, we chose high mobility group box 1 (HMGB1), osteopontin (OPN), and hyaluronic acid (HA), all three of them have been well documented to be involved in endometriosis and fibrosis, as potential biomarkers. In addition, we performed immunohistochemistry analysis of HMGB1, OPN, and the receptors for HMGB1, such as toll-like receptor 4 (TLR4), nuclear factor κB (NF-κB), proliferating cell nuclear antigen (PCNA), interleukin-33 (IL-33), and receptor for advanced glycation endproducts (RAGE)-a pattern recognition receptor, with HMGB1 being its important ligand. We then evaluated the same set of putative markers in 30 women with ovarian endometriomas and 20 without endometriosis, and reevaluated the 3 plasma markers 3 months after the surgical removal of all visible endometriotic lesions. In mouse, the lesional staining levels of OPN, RAGE, and IL-33 were all significantly higher than that of normal endometrium, and increased progressively as lesions progressed. In contrast to HMGB1, TLR4, p-p65 and PCNA staining levels were decreased progressively. In humans, lesional staining levels of OPN correlated positively, while that of HMGB1 correlated negatively with the extent of fibrosis. All three plasma markers correlated positively with the extent of lesional fibrosis. Through this integrated approach, we identified plasma HMGB1, OPN and HA as promising admissible biomarkers for endometriosis.

Figure 1

Summary results of the mouse experiment. (A) The dynamic changes of the mean bodyweight in different groups of mice. Data are presented as mean ± SD. (B) Kinetics of the mean hotplate latency in different groups. Data are presented as mean ± SD. (C) Boxplot of lesion weight in different groups. (D) Boxplot of the plasma HMGB1 levels (in pg/mL). (E) Boxplot of the plasma OPN levels (in ng/mL). (F) Boxplot of the plasma HA levels (in ng/mL). For all three groups, n = 8. Symbols of statistical significance levels: *or ^#p < 0.05; **or ^##p < 0.01; ***p < 0.001; NS: not statistically significant, i.e. p > 0.05, all by Wilcoxon’s rank test in reference to the control group (in black) or the ENDO2 group (in red) (except in (A,B), where the comparison was made by Kruskal’s test,). (G) Representative microphotographs of OPN, RAGE, IL-33, HMGB1, TLR4, p-p65, PCNA and Masson staining in CTRL mice and ENDO mice after 2 and 5 weeks of induced endometriosis. Bar = 50 µm.

Figure 2

Summary immunostaining results of the mouse experiment. Boxplot of lesional OPN (A), RAGE (B), IL-33 (C), HMGB1 (D), TLR4 (E), p-p65 (F), PCNA (G) staining in the epithelial component, and the extent of lesional fibrosis (G) in different groups. For all three groups, n = 8. Symbols of statistical significance levels: *or ^#p < 0.05; **or ^##p < 0.01; ***p < 0.001; NS: not statistically significant, i.e. p > 0.05, all by Wilcoxon’s rank test in reference to the control group (in black) or the ENDO2 group (in red).

Figure 3

Relationship between the extent of lesional fibrosis and various factors. Scatter plot showing the relationship between the extent of lesional fibrosis and lesion weight (A), hotplate latency (B), plasma OPN levels (C), plasma HA levels (D), lesional staining levels of OPN (E), RAGE (F), IL-33 (G), PCNA (H), and TLR4 (I). (J) The scatter plot showing the relationship between the plasma OPN levels and lesional OPN staining levels. The dashed line represents the linear regression line, and the number is the Pearson’s correlation coefficient, followed by the symbols of statistical significance levels: *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 4

Results of multidimensional scaling analysis. Results of multidimensional scaling analysis using variables of hotplate latency (LTCY) and HMGB1 (A), OPN (B), and HA (C). Each character in the figure represents one data point from one mouse, with CT, E2, and E5 representing mouse from the control, ENDO2, and ENDO5 groups, respectively. Different groups of mice are all correctly classified.

Figure 5

Summary of human data. Data summary of the plasma HMGB1 (A), OPN (B), and HA (C) levels in the peripheral blood from women without endometriosis (Control), women with ovarian endometriomas before the lesional-removal surgery (Endo-pre-op) and 3 months after the surgery (Endo-3M post-op). The statistical significance levels of the between-group difference in the designated groups are shown. Scatter plot showing the relationship between the rASRM scores and plasma OPN (D), HMGB1 (E), and HA (F). Scatter plot showing the relationship between the extent of lesional fibrosis and plasma OPN (G), HMGB1 (H), and HA (I). In (D–H), The dashed line represents the linear regression line, and the number is the Pearson’s correlation coefficient, followed by the symbols of statistical significance levels. Each dot in the figure represents one data point/patient, with the red and brown dots indicating the patient had unilateral and bilaterial ovarian endometriomas. Levels of statistical significance are ***p < 0.001; NS, nonsignificant, i.e. p > 0.05.

Figure 6

Representative microphotographs of immunohistochemical staining by the extent of lesional fibrosis. Representative microphotographs of lesional staining of OPN, RAGE, IL-33, HMGB1, TLR4, p-p65, and PCNA, and Masson staining from patients with endometriosis. Bar = 50 µm.

Figure 7

Relationship between the extent of lesional fibrosis and various factors. Scatter plot showing the relationship between the extent of lesional fibrosis and lesional staining of OPN (A), RAGE (B), IL-33 (C), HMGB1 (D), TLR4 (E), p-p65 (F), and PCNA (G) in the epithelial component, and plasma levels of OPN (H) and HMGB1 (I). The dashed line represents the linear regression line, and the number is the Pearson’s correlation coefficient, followed by the symbols of statistical significance levels: ***p < 0.001. Each dot in the figure represents one data point/patient, with the red and brown dots indicating the patient had unilateral and bilateral ovarian endometriomas.

Figure 8

Receiver operating characteristic (ROC) curves for individual markers and their select combinations. Different ROC curves are represented by curves of different colors. The dashed diagonal line represents the ROC curve under a random classifier.

Figure 9

Results of multidimensional scaling analysis of human data. Results of multidimensional scaling analysis using variables of the severity of dysmenorrhea (DYS), OPN, and HA. Each character in the figure represents one data point from one patient, with E and C representing patient with endometriosis and without, respectively.