Immunomodulatory Effects of Metformin Treatment in Pregnant Women With PCOS

Abstract

Context: Polycystic ovary syndrome (PCOS) is a common endocrine disorder associated with low-grade systemic inflammation and increased risk of pregnancy complications. Metformin treatment reduces the risk of late miscarriage and preterm birth in pregnant women with PCOS. Whether the protective effect of metformin involves immunological changes has not been determined.

Objective: To investigate the effect of metformin on the maternal immunological status in women with PCOS.

Methods: A post-hoc analysis was performed of two randomized controlled trials, PregMet and PregMet2, including longitudinal maternal serum samples from 615 women with PCOS. Women were randomized to metformin or placebo from first trimester to delivery. Twenty-two cytokines and C-reactive protein were measured in serum sampled at gestational weeks 5 to 12, 19, 32, and 36.

Results: Metformin treatment was associated with higher serum levels of several multifunctional cytokines throughout pregnancy, with the strongest effect on eotaxin (P < .001), interleukin-17 (P = .03), and basic fibroblast growth factor (P = .04). Assessment of the combined cytokine development confirmed the impact of metformin on half of the 22 cytokines. The immunomodulating effect of metformin was more potent in normal weight and overweight women than in obese women. Moreover, normoandrogenic women had the strongest effect of metformin in early pregnancy, whereas hyperandrogenic women presented increasing effect throughout pregnancy.

Conclusion: It appears that metformin has immunomodulating rather than anti-inflammatory properties in pregnancy. Its effect on the serum levels of many multifunctional cytokines demonstrates robust, persisting, and body mass-dependent immune mobilization in pregnant women with PCOS.

Keywords: PCOS; cytokine; immunology; metformin; multivariate analysis; pregnancy.

© The Author(s) 2023. Published by Oxford University Press on behalf of the Endocrine Society.

Figures

Figure 1.

Development of cytokines and CRP expressed as β-coefficients from LMM in the metformin group and the placebo group. P values were adjusted for multiple testing using the Benjamini-Hochberg procedure. Filled circles represent either a significant development over time in the placebo group or a significantly different development in the metformin group compared to the placebo group. Abbreviations: CRP, C-reactive protein; FGF-b, basic fibroblast growth factor; G-CSF, granulocyte colony-stimulating factor; GM, granulocyte macrophage; IP, interferon-γ–induced protein; LMM, linear mixed model; MCP, monocyte chemotactic protein; MIP, macrophage inflammatory protein; PDGF, platelet-derived growth factor; Ra, receptor antagonist.

Figure 2.

Combined assessment and comparison of cytokine-development over time by RM-ASCA+ in women receiving metformin or placebo throughout pregnancy. (A) Development of cytokine levels in women receiving placebo throughout pregnancy. Each circle represents the score on the first principal component (PC1), which explains most of the variation in the dataset. This score must be interpreted with the loading plot for cytokines (right), explaining how much each variable is contributing to the scores. Variables that have positive loadings follow the time curve, whereas variables that have negative loadings have an inverse development over time. A high loading, either negative or positive, means that this cytokine is important to describe the development over time. Each score and loading have a bar representing the uncertainty. If the uncertainty of two scores overlaps, the difference between them was not considered robust, and if the uncertainty of loadings touches the zero-line, their contribution to the score plot was not considered robust. (B) Differences in cytokine development between metformin- and placebo-treated women throughout pregnancy. The development over time in the placebo group is flattened for easier interpretation. Cytokines with positive loadings are higher in the metformin group, whereas cytokines with negative loadings are lower in the metformin group. A high loading, either negative or positive, means that this cytokine is important to describe the difference in development between metformin and placebo groups. Abbreviations: CRP, C-reactive protein; FGF-b, basic fibroblast growth factor; G-CSF, granulocyte colony-stimulating factor; GM, granulocyte macrophage; IP, interferon-γ–induced protein; MCP, monocyte chemotactic protein; MIP, macrophage inflammatory protein; PC, principal component; PDGF, platelet-derived growth factor; Ra, receptor antagonist.

Figure 3.

The effect of metformin treatment in different BMI groups by RM-ASCA+. (A) Difference in cytokine development in normal weight women receiving metformin or placebo throughout pregnancy (n = 213). (B) Difference in cytokine development in overweight women receiving metformin or placebo throughout pregnancy (n = 175). (C) Difference in cytokine development in obese women receiving metformin or placebo throughout pregnancy (n = 227). The development over time in the placebo group is flattened for easier interpretation, and the development curves of the placebo groups are found in Supplementary Fig. S2 (46). The left panel models the score on PC1 and must be interpreted with the loading plot for cytokines (right), explaining how much each variable is contributing to the scores. Each score and loading have a bar representing the uncertainty. If the uncertainty of two scores overlaps, the difference between them was not considered robust, and if the uncertainty of loadings touches the zero-line, their contribution to the score plot was not considered robust. Cytokines with positive loadings are higher in the metformin group, whereas cytokines with negative loadings are lower in the metformin group. A high loading, either negative or positive, means that this cytokine is important to describe the difference in development between metformin and placebo group. Abbreviations: BMI, body mass index; CRP, C-reactive protein; FGF-b, basic fibroblast growth factor; G-CSF, granulocyte colony stimulating factor; GM, granulocyte macrophage; IP, interferon-γ–induced protein; MCP, monocyte chemotactic protein; MIP, macrophage inflammatory protein; PC, principal component; PDGF, platelet-derived growth factor; Ra, receptor antagonist.

Figure 4.

The effect of metformin treatment in hyperandrogenic and normoandrogenic women. (A) Development of cytokine levels in hyperandrogenic women receiving metformin or placebo throughout pregnancy (n = 467). (B) Development of cytokine levels in normoandrogenic women receiving metformin or placebo throughout pregnancy (n = 144). The development over time in the placebo group is flattened for easier interpretation, and the development curves of the placebo groups are found in Supplementary Fig. S4 (46). The left panel models the score on PC1 and must be interpreted with the loading plot for cytokines (right), explaining how much each variable is contributing to the scores. Each score and loading have a bar representing the uncertainty. If the uncertainty of two scores overlaps, the difference between them was not considered robust, and if the uncertainty of loadings touches the zero-line, their contribution to the score plot was not considered robust. Cytokines with positive loadings have higher levels in the metformin group, whereas cytokines with negative loadings have lower levels in the metformin group. A high loading, either negative or positive, means that this cytokine is important to describe the difference in development between the metformin group and the placebo group. Abbreviations: CRP, C-reactive protein; FGF-b, basic fibroblast growth factor; G-CSF, granulocyte colony stimulating factor; GM, granulocyte macrophage; IP, interferon-γ-induced protein; MCP, monocyte chemotactic protein; MIP, macrophage inflammatory protein; PC, principal component; PDGF, platelet-derived growth factor; Ra, receptor antagonist.