Summary

Graphical abstract

Open in a separate window

Introduction

Glycosylation of immunoglobulin G (IgG) is an important regulator of the immune system (Martinić Kavur et al., 2021). In addition to directly affecting the effector functions of IgG by promoting its binding to different Fc receptors (Nimmerjahn and Ravetch, 2008a), glycans attached to IgG have numerous other roles in the regulation of the immune system (Ipsen-Escobedo and Nimmerjahn, 2018; Seeling et al., 2017). Glycans are also important mediators of the anti-inflammatory activity of IgG (Karsten et al., 2012; Schwab and Nimmerjahn, 2013) as well as the driving force for one of the molecular mechanisms underlying immunosuppressive activity of intravenous IgG (IVIG) therapy used to treat a variety of immunological disorders (Nimmerjahn and Ravetch, 2008b). Changes in IgG glycosylation are also believed to be an important contributor to aging at the molecular level through a process called inflammaging (Franceschi et al., 2018). Furthermore, hyposialylated IgG, which associates with ageing, was shown to contribute to the development of hypertension in animal models through FcγRII on endothelial cells (Peng et al., 2019; Sundgren et al., 2015).

IgG glycome composition is tightly regulated and, despite the absence of a direct genetic template, is heritable to a significant extent (Menni et al., 2013; Pučić et al., 2011). Alternative glycosylation (the attachment of different glycan structures to the same glycosylation site) is structurally and functionally analogous to coding mutations, but instead of being defined by sequence variation in individual genes, glycome composition is inherited as a complex trait encoded by multiple genes (Krištić et al., 2018). Genome-wide association studies identified a network of over 30 genes that associate with the IgG glycome composition (Klarić et al., 2020; Lauc et al., 2013; Shen et al., 2017). One of the regulators of IgG glycosylation are sex hormones, and the association between estradiol and the IgG glycome composition was confirmed to be causal (Ercan et al., 2017). A recent randomized placebo-controlled clinical study demonstrated that the deprivation of gonadal hormones resulted in an increase of biological age measured by IgG glycans, which was completely prevented by estradiol supplementation (Juric et al., 2020).

Large population studies revealed that IgG glycome composition changes with age and that in females this change is particularly pronounced in the time preceding the average age of menopause (Krištić et al., 2014; Stambuk et al., 2020). However, the association between changes in the IgG glycome composition and menopause has not yet been validated. Perimenopause is accompanied by different symptoms and is very often misdiagnosed due to changing hormone levels and symptom overlap with other conditions. Menopause associates with an increased risk of developing a range of diseases affecting metabolism, bones, cardiovascular system, brain, and even cancer (Lobo et al., 2014). Early perimenopause diagnosis would enable timely interventions to ease the symptoms, decrease risks of developing accompanying diseases, and timely monitoring of any changes in women's health status due to hormonal changes. To fill this knowledge gap and explore the association of IgG glycome changes with perimenopause, we analyzed IgG N-glycome composition in 5,080 samples from 1,940 females in multiple time points during their transition from premenopause to postmenopause, along with 274 samples from 113 males. As an additional control, samples from 301 females in an independent validation cohort were analyzed.

Results

IgG glycome composition in premenopause and menopause

IgG glycome composition was analyzed in multiple samples from 1,940 females and 113 males from the UK registry of adult twins. The 1,087 samples from 732 individual twins (500 families) were collected prior to the onset of menopause, whereas 3,993 samples from 1,678 individual twins (969 families) were collected after the onset of menopause (Table S1). In addition, IgG glycome was analyzed in 113 men (274 samples) as an additional control. A representative chromatogram with detailed structures of individual IgG glycans is shown in Figure S1. Mixed modelling was used to determine average levels of individual IgG glycans in premenopausal women, menopausal women, and men. Large and statistically significant differences were observed in multiple glycans, mainly reflecting decreased galactosylation and sialylation of glycans in menopausal women (Figure 1). Levels of individual IgG glycans are available in Figure S2.

Open in a separate window

Figure 1

IgG glycome composition in premenopausal women, menopausal women, and men

Mean glycan abundances (% of total IgG glycome) and corresponding 95% confidence intervals were estimated from the fitted mixed model with logit transformed glycan as dependent variable and sex, menopausal status (nested within sex), age, and age-menopausal status interaction as fixed factors, as well as family ID and individual ID (nested within family ID) as random intercepts and age as random slope. Only adjusted post-hoc p values less than 0.1 are shown. B, bisecting N-acetylglucosamine (GlcNAc); CF, core fucose; G0, agalactosylated glycans; G1, monogalactosylated glycans; G2, digalactosylated glycans; S1, monosialylated glycans; and S2, disialylated glycans.

Average yearly change of IgG glycans in perimenopause

It is known that IgG galactosylation and sialylation decrease with age (Krištić et al., 2014), thus we attempted to evaluate whether the changes in glycans associated with the transition to menopause were more extensive than the age-related changes. Because we had multiple samples from the same individual, we were able to calculate the rate of change of individual IgG glycans over time. A subset of women in our cohort (n = 379) entered menopause between two sampling time points, thus for them, we were also able to calculate the rate of changes in individual IgG glycans during the perimenopausal period. The comparison of the rate of age-related IgG glycome changes (when there was no change of the menopause status) with the rate of IgG glycome changes during perimenopause (i.e., between time points when the transition to menopause occurred) revealed statistically significant differences for a number of glycans (Figure S3). The most prominent differences in the perimenopause period were the significantly higher rates of increase in the agalactosylated structures (G0) and the decrease of digalactosylated (G2) and monosialylated (S1) glycans (Figure 2). During perimenopause women had a significant higher rate of increase in agalactosylated structures (0.051/yr; 95%CI = 0.043–0.059, p < 0.001) and decrease in digalactosylated (−0.043/yr; 95%CI = −0.050 to −0.037, p < 0.001) and monosialylated glycans (−0.029/yr; 95%CI = −0.034 to −0.024, p < 0.001), compared with premenopausal women.

Open in a separate window

Figure 2

Average yearly change of IgG glycans in females during the perimenopause period, females in pre- or postmenopause, and males

Mean yearly change in IgG glycan abundances and corresponding 95% confidence intervals were estimated from the fitted mixed model with logit transformed glycan as dependent variable and sex, menopausal status (nested within sex), age and age-menopausal status interaction as fixed factors, as well as family ID and individual ID (nested within family ID) as random intercepts and age as random slope. Only adjusted post-hoc p values less than 0.1 are shown. B, bisecting GlcNAc; CF, core fucose; G0, agalactosylated glycans; G1, monogalactosylated glycans; G2, digalactosylated glycans; S1, monosialylated glycans; and S2, disialylated glycans.

Prediction of perimenopause using IgG glycans

Using single-point measurements of IgG N-glycans we attempted to predict menopause status. The lasso model was fitted with menopausal status (TRUE for those who entered menopause and FALSE for all others) as dependent variable and age and all directly measured IgG glycans as predictors. The model was fitted on a randomly chosen one twin per family (training set) in the subset of women with a two time point median age between 45 and 55 years. The remaining twins were used for testing the fitted model. The same train and test subset model with only mean age as a predictor were used to fit and test the model for comparison (Figure 3A). IgG N-glycome in combination with age measured in a single time point has shown a good classification performance, with the AUC of 0.853 (95% CI 0.823–0.880) outperforming only age as a predictor of perimenopause (AUC of 0.818, 95% CI 0.783–0.849). Because changes in the IgG glycome composition were the most prominent in the period of perimenopause, we used the rate of changes in glycans to predict whether given women already entered the period of perimenopause or not (Figure 3B). The addition of the information about changes in the IgG glycan levels to a prediction model notably increases AUC value compared with using only age (0.694, 95% CI 0.637–0.751 compared with 0.507, 95% CI 0.443–0.570, respectively).

Open in a separate window

Figure 3

Receiver operating characteristic (ROC) curves for the prediction of perimenopause

IgG glycans measured in a single time point (A) and the change in IgG glycan levels between the two time points (B). Graphs are showing the area under the curve (AUC) values with 95% confidence intervals.

IgG glycome composition in replication cohorts

The sampling in replication cohorts was cross-sectional, and there was only one time point for each sample, which did not allow to infer any causal relationships such as monitoring changes in IgG N-glycome over time. Because of that limitation, this was not a perfect replication cohort for our model validation; therefore, it was not possible to assess whether someone is perimenopausal or not. What we could replicate are the consistent differences between premenopausal and menopausal females of any age between 45 and 55 years (Figure 4).

Open in a separate window

Figure 4

IgG glycome composition in premenopausal and menopausal women of any age between 45 and 55 years in replication cohorts

Glycan abundance (%) is presented as % of total IgG glycome

Lines are representing mean levels of glycan abundance by age for each cohort with SE of mean presented as the shaded area around the mean. B, bisecting GlcNAc; CF, core fucose; G0, agalactosylated glycans; G1, monogalactosylated glycans; G2, digalactosylated glycans; S1, monosialylated glycans; and S2, disialylated glycans.

Discussion

By analyzing IgG glycome in multiple samples from the same individuals, we have shown the association between the period of perimenopause and the extensive changes in the IgG glycome composition. Aging associates with the transition of the IgG glycome from an inflammation-suppressive to a proinflammatory composition, but this change is much faster in the period when women are progressing from a regular cycle to menopause, as was shown in several population studies (Krištić et al., 2014; Stambuk et al., 2020). The most extensive change in the IgG glycome during perimenopause was a decrease in galactosylation and sialylation, a change that is generally associated with aging (Martinić Kavur et al., 2021). The depletion of estrogen during perimenopause and menopause is an important component of aging, as low hormone levels increase vulnerability to diseases in hormone-responsive tissues, including the brain, bone, and the cardiovascular system (Lobo et al., 2014). Menopause is, therefore, more complex than just being a natural condition resulting in symptoms; it needs to be considered as a long-term hormone deficiency potentially leading to higher health risks if not appropriately managed with lifestyle adjustments or hormone replacement therapy (HRT) in consonance with a gynecologist and/or general practitioner (Gambacciani et al., 2019).

The age-related gradual decrease in the level of galactosylated IgG and the increase in agalactosylated IgG is known to exacerbate low-grade chronic inflammation (De Martinis et al., 2005; Franceschi et al., 2000; Monti et al., 2017) and acts as an effector of proinflammatory pathological changes (Dall’Olio et al., 2013; Franceschi et al., 2007). Decreased levels of IgG galactosylation and/or sialylation have been associated with a wide range of autoimmune conditions, such as systemic lupus erythematosus, inflammatory bowel disease, and many others. Changes in these IgG glycosylation traits are also associated with disease progression, disease activity, symptom severity, and response to treatment. In many cases, these changes occur before the onset of symptoms (Gudelj et al., 2018a).

The rate of change from “young” to “old” IgG glycome increased significantly in the perimenopausal period, suggesting that the loss of immunosuppressive functions of IgG may be an underlying molecular mediator of at least some health risks in this period. The results of this study also indicate the potential of the IgG glycome composition as a biomarker for perimenopause. Perimenopause can last up to 15 years and is difficult to diagnose due to highly irregular hormonal cycles (National Institute for Health and Care Excellence, 2015). As a result of poor awareness and inappropriate use of hormonal tests, women are often misdiagnosed with conditions such as fibromyalgia, migraines, depression, or chronic fatigue syndrome and are frequently prescribed antidepressants despite there being no evidence to support their use to improve the low mood associated with perimenopause or menopause (Leonhardt, 2019). The importance of early diagnosis of perimenopause is supported by the research that has shown that early use of HRT leads to a greater reduction of future disease incidence as well as minimizing the duration of symptoms (Ward and Deneris, 2018).

Recent placebo-controlled randomized trials demonstrated the importance of estrogens in the regulation of IgG glycosylation (Ercan et al., 2017; Juric et al., 2020), thus it is reasonable to assume that the decrease in estrogen levels is an important driver of changes in IgG glycosylation in perimenopause. Contrary to estrogens that have a very short half-life in circulation, IgG has a half-life of 3 weeks (Vidarsson et al., 2014), which implies that the current composition of the IgG glycome is affected by average concentrations of estrogen in the past 3–4 weeks. Because changes in the IgG glycome during perimenopause are so extensive, they could potentially be used as a diagnostic tool. In fact, IgG glycome composition could potentially be used to determine long-term average concentrations of sex hormones in the same way glycated hemoglobin (HbA1c) is used to determine long-term concentrations of blood glucose.

Conclusions

Changes in IgG glycosylation from “young” to “old” glycome associate with many health risks that accompany menopause (Martinić Kavur et al., 2021). In some diseases such as rheumatoid arthritis (Ercan et al., 2010; Gudelj et al., 2018b) and cardiovascular diseases (Kifer et al., 2021; Menni et al., 2018; Wittenbecher et al., 2020), this change was shown to occur years before disease onset, suggesting the causal role of IgG glycans in disease development. Proinflammatory IgG glycome and the resulting decrease in the ability of IgG to suppress low-grade chronic inflammation may be an important molecular mechanism that mediates the increased health risk in perimenopause, thus this topic should be studied in more detail.

STARMethods

Acknowledgements

Notes

Published: March 18, 2022

Footnotes

References