Macroheterogeneity relates to the presence or absence of a oligosaccharide chain at any of the two known N-glycosylation sites (site occupancy) [17] in the beta subunit. In vivo, the macroheterogeneity of circulating hFSH is dynamic and may have a physiologic role. In women of reproductive age, more glycosylated (fully glycosylated) and acidic glycoforms (with prolonged in vivo half-life due to reduced renal clearance) are secreted during the early and mid-follicular phases, compared with less sialylated, glycosylated (hypo-glycosylated) glycoforms, which are more predominant before ovulation [32,33]. Highly acidic isoforms are more predominant after the menopause than during the fertile lifespan [34]. Furthermore, tri-glycosylated hFSH (hFSH18/21) is more abundant in young women, whereas tetra- glycosylated (hFSH24) and highly sialylated forms are more abundant in peri/postmenopausal women [32,35]. Glycosylation at αAsn 52 is essental for FSH bioactivity, as it has an important role in the assembly of the functional FSH heterodimer and its subsequent stability; it also has a pivotal role in FSH receptor (FSHR) activation and signalling, whereby glycoforms with smaller and more compact glycosylation at αAsn 52 can fit into the central cavity of the FSHR more rapidly than bulkier and extended glycans, leading to a more rapid response [17].

Microheterogeneity relates to the structural variation in the type of carbohydrates comprising the oligosaccharide chains attached to the protein core and the branching of these chains (antennarity) into bi-, tri- and tetra-antennary structures [36,37,38]. The building block N-acetyl glucosamine (GlcNAc) is linked to asparagine followed by the addition of another GlcNAc, then by one to three Mannose residues that can branch into 1 to 4 antennae. The antennae are then extended by GlcNAc and galactose, the latter of which can be capped by N-acetyl neuraminic acid (sialic acid) [30]. The average number of antennae per glycan is reflected by the A-Index, which is calculated for each N-glycosylation site and as a mean over the entire molecule (see Section 4).

Sialylation relates to the inclusion of sialic acid in the glycoprotein antennae. The addition of a sialic acid cap imparts a negative charge on each antenna [30]. The degree of sialic acid capping can vary among glycoforms, whereby variants with a high degree of sialylation are more acidic than those with low sialic acid content [36,37,38]. A high level of sialylation, in combination with the presence of bulkier glycans, may contribute to a longer half-life through reduced glomerular filtration and, therefore, higher net in vivo potency. CHO cells, in which r-hFSH is produced, do not have the ability to synthesize sialic acid attached in the position α2-6, so only α2-3 sialic acid is found in reference r-hFSH-alfa and biosimilar preparations [30]. The average number of sialic acid moieties per glycan is reflected by the S-Index, which is calculated for each N-glycosylation site and as a mean over the entire molecule (see Section 4). The NGNA index reflects the content of glycans with N-glyconeuramic acid (NGNA) moieties. NGNA may be linked to immunogenic reactions, as humans do not produce CMP-N-acetylneuraminic acid hydroxylase, the enzyme responsible for this glycan modification [38]. Anti-NGNA activity has been reported in 85% of healthy humans, suggesting its potential for eliciting an immune response in humans [39]. The NGNA index is calculated for each N-glycosylation site and as a mean over the entire molecule (see Section 4).

O-acetylation: The nine-carbon backbone of sialic acids can undergo extensive enzymatic modification in nature, and O-acetylation at the C-4/7/8/9 positions in particular is widely observed [40]. O-acetylation increases the hydrophobic character of sialylated glycans and can change the biophysical properties of the glycoprotein, potentially leading to changes in activity and glycan antigen recognition [40].

Post-translational modifications can include the oxidation of methionine residues. The methionine residues in FSH are not directly located in regions that are critical for binding to the FSHR: methionine 29 is involved in α–β subunit heterodimerization; methionine 47 is located close to the FSHR binding site, but is not directly involved in ligand–receptor interaction; methionine 71 is located close to the heterodimerization site but is not directly involved in heterodimerization; and methionine 109 is located in the non-structural C-terminal region. However, the oxidation of these residues may lead to conformational changes, with the potential for indirect effects on biological activity, pharmacokinetics or protein aggregation, and the alteration of the immunogenicity of therapeutic proteins [41]. In contrast to the post-translational modifications in the oligosaccharide chains attached to the protein core, other modifications, such as the conversion of asparagine to succinimide and N-terminal clipping, are not known to have an impact on biological activity.

Higher-order structure refers to the self-assembly into either the seconday-, tertiary- and quarternary-order structure of a protein. Higher-order structure is responsible for the correct folding and three-dimensional shape of a protein and is strongly dependent on the protein environment; therefore, different formulations can bear conformational differences compared with the reference preparation, which can have an impact on the activity of the molecule.