What a Glycan Age Test Measures

A glycan age test measures the sugar structures attached to immunoglobulin G (IgG), your most abundant circulating antibody, and converts a handful of ratios among those structures into a number expressed in years. The biology underneath is real and well characterized: the N-glycan attached at asparagine 297 of the IgG Fc region loses terminal galactose and sialic acid with age, and that shift changes how the antibody engages Fc receptors and complement, tilting IgG toward pro-inflammatory effector function ¹. The number on the report is a supervised model trained to predict chronological age from those ratios, so it inherits both the signal of the glycome and the limitations of any single-modality age predictor. Treat the underlying glycan measurements as the useful object and the age-in-years translation as a convenient rescaling.

What the assay physically measures

IgG carries a single conserved N-glycosylation site per heavy chain. The glycan there is a biantennary complex structure built on a core of two N-acetylglucosamine and three mannose residues, and it varies in four ways that matter: core fucosylation, bisecting N-acetylglucosamine, the number of terminal galactose residues (zero, one, or two, giving the G0/G1/G2 nomenclature), and terminal sialic acid.

Two analytical routes dominate. The first releases glycans enzymatically with PNGase F, labels them with a fluorophore such as 2-aminobenzamide (2-AB) or procainamide, and separates them by hydrophilic interaction ultra-performance liquid chromatography (HILIC-UPLC) with fluorescence detection. The output is a chromatogram integrated into a fixed set of peaks, conventionally 24 for IgG (GP1 through GP24), each peak containing one or a few known structures. The second route uses liquid chromatography–mass spectrometry on tryptic glycopeptides, which preserves IgG subclass information because the tryptic peptide around Asn297 differs between IgG1, IgG2/3, and IgG4. Subclass-resolved data is more informative and more work to process. If you have the choice and intend to analyze the data yourself, take the LC-MS glycopeptide route.

The age-associated pattern is consistent across populations: agalactosylated structures (G0F, G0FN) rise, digalactosylated and sialylated structures fall, and the shift accelerates in women around menopause. This is not a passive readout of age. Removing galactose and sialic acid from Fc glycans increases affinity for FcγRIII and for complement activation through the lectin pathway, which is why the IgG glycome is described as both a biomarker and a functional effector of aging and mortality risk ². Glycan changes on immune cells themselves also alter function: age-related remodeling of sialoglycans has been shown to dampen CD8+ T cell responses in mice ³, and blocking glycan degradation attenuated inflammaging and cognitive impairment in an HIV model ⁴.

How the years number is produced

The transformation from peaks to years is a regression. A vendor fits a model (elastic net, or a small set of derived traits in a linear or spline model) predicting chronological age from normalized glycan features in a training cohort, then applies it to you. Two consequences follow directly from that construction.

First, prediction error in the training data sets a floor on interpretability. Glycan predictors of chronological age typically leave a residual on the order of several years, which means a reported age within roughly a decade of your calendar age is consistent with an unremarkable profile. A single point estimate should not be read to two significant figures.

Second, the model is trained on chronological age, so what it captures best is the average trajectory of the population, and the deviation from that trajectory is the part carrying information. Epigenetic clocks have the same structure and the same critique: predictors trained on age recover age-correlated variance, while predictors trained on mortality or morbidity capture different and often more clinically aligned signal ⁵. The glycome’s specific claim is that its residual tracks chronic low-grade inflammation, the process described as inflammaging that links immune activation to cardiovascular, metabolic, and neurodegenerative disease ⁶. That connection is mechanistically plausible and supported, and it is also why a glycan score and a methylation score can disagree without either being wrong. They are measuring different layers.

Working with the raw data

Ask the lab for the integrated peak table, not the report. What you want is a matrix of samples by peaks with raw areas, plus the structure annotation for each peak and the plate and run identifiers. Without plate metadata you cannot correct batch effects, and batch effects in glycomics are large enough to swamp a year of biological change.

The processing sequence we would use is: total area normalization (divide each peak by the summed area of all peaks in that sample, giving relative abundance), then a log or logit transform to stabilize variance, then batch correction with ComBat (sva::ComBat in R) using plate as batch. The glycanr package on CRAN implements the standard normalizations (tanorm, quantilenorm, medianquotientnorm) directly on long-format glycan data frames and will save you writing them. Only after normalization should you compute derived traits, which are sums and ratios over peak groups: total galactosylation as the abundance-weighted mean of G0, G1, G2 fractions, sialylation per galactose, bisecting GlcNAc fraction, core fucosylation fraction. The mapping from peak index to structure is method-specific, so take it from the lab’s annotation file rather than assuming GP numbering matches another paper’s.

Failure modes worth knowing. Sialic acid is labile under acidic or high-temperature conditions, so poor sample handling systematically deflates sialylation and inflates apparent age. Serum and plasma are not interchangeable across a longitudinal series. Affinity purification of IgG that carries albumin or IgA contamination introduces non-IgG glycans into early-eluting peaks. And because the score is noisy, a repeat test after an intervention will regress toward the mean, which is the single most common way people over-read an improvement.

Using it longitudinally

The strongest use of a glycan measurement is as a repeated measure on yourself with consistent methodology, ideally with samples from the same intervention window run on the same plate. IgG glycosylation responds to state: it is regulated in part epigenetically through glycosyltransferase gene expression, which is one reason it can shift over months rather than being fixed at birth ⁷. That responsiveness is the property that makes it interesting and also the property that makes single measurements unstable.

Interpret the score alongside markers that report the same underlying process from other angles: high-sensitivity C-reactive protein, IL-6, and inflammation-weighted proteomic panels, since vascular and metabolic aging share mechanisms with immune aging ⁸. If any of those are abnormal, or if a glycan profile shifts sharply without an explanation, that is a conversation for a physician, not a self-directed project. Nothing here supports a diagnosis or an intervention.

Questions people also ask

Is GlycanAge legit? The measurement is legitimate laboratory glycomics, and the biology connecting IgG galactosylation to inflammation and mortality risk is supported by substantial literature ². The translation into a single age in years is a modeling choice with real uncertainty, and the marketing claim that it uniquely tracks lifestyle response goes beyond what published validation establishes. Buy it for the glycan data, not the headline number.

What are the four types of age? The common framing is chronological age (time since birth), biological age (a molecular estimate of functional state), psychological age, and social age. In molecular work only the first two are measurable, and biological age is a family of estimates rather than one quantity.

What are the different types of biological age clocks? Epigenetic clocks built on DNA methylation ⁵, transcriptomic clocks from RNA sequencing, proteomic clocks from plasma protein panels, glycan clocks from IgG N-glycans, and image-based clocks derived from nuclear morphology and chromatin organization ⁹. They correlate imperfectly with one another because each samples a different layer of regulation.

Can a glycan profile change? Yes. IgG glycosylation shifts over months in response to metabolic and hormonal state, and it is a stated target for intervention research rather than a fixed trait ¹. Whether a given change improves outcomes is not established.

Oak builds longitudinal molecular profiles of individuals: whole-genome sequencing, RNA sequencing, proteomics, blood biomarkers, and continuous glucose data, integrated into one model of you. Build your profile.

Footnotes

1. Gordan Lauc. Can we suppress chronic systemic inflammation and postpone age-related diseases by targeting the IgG glycome?. Expert Opinion on Therapeutic Targets, 2023. https://doi.org/10.1080/14728222.2023.2277218 ↩ ↩²

2. Gordan Lauc, Anika Mijakovac, Elena Butz, et al. The Immunoglobulin G Glycome: A Modifiable Biomarker and Functional Effector of Aging, Disease, and Mortality. 2026. https://doi.org/10.21203/rs.3.rs-9488315/v1 ↩ ↩²

3. Hanlin Zhang, C. Kimberly Tsui, Jesse Garcia Castillo, et al. Age-related remodeling of the sialoglycans dampens murine CD8 + T cell function. Science Advances, 2025. https://doi.org/10.1126/sciadv.adw6755 ↩

4. Leila B. Giron, Alejandra Borjabad, Eran Hadas, et al. Inhibiting glycan degradation prevents HIV-induced inflammaging and cognitive impairment. Med, 2026. https://doi.org/10.1016/j.medj.2026.101175 ↩

5. Adam Li, Zane Koch, Trey Ideker. Epigenetic aging: Biological age prediction and informing a mechanistic theory of aging. Journal of Internal Medicine, 2022. https://doi.org/10.1111/joim.13533 ↩ ↩²

6. Claudio Franceschi, Paolo Garagnani, Paolo Parini, et al. Inflammaging: a new immune–metabolic viewpoint for age-related diseases. Nature Reviews Endocrinology, 2018. https://doi.org/10.1038/s41574-018-0059-4 ↩

7. Vlatka Zoldoš, Srđana Grgurević, Gordan Lauc. Epigenetic regulation of protein glycosylation. bmc, 2010. https://doi.org/10.1515/bmc.2010.027 ↩

8. Yalan Wu, Xiao Yu Tian. Panvascular aging as a unifying framework: convergent mechanisms across age-related diseases. Vessel Plus, 2026. https://doi.org/10.20517/2574-1209.2025.120 ↩

9. Martin Alvarez-Kuglen, Kenta Ninomiya, Haodong Qin, et al. ImAge quantitates aging and rejuvenation. Nature Aging, 2024. https://doi.org/10.1038/s43587-024-00685-1 ↩