The synthetic process scheme of MGA was compared with published reaction schemes for GA and patents describing its synthesis (Konfino et al., 2007; Teitelbaum et al., 1971). The amino acids that comprise GA are the levorotatory isomers of glutamic acid, lysine, alanine, and tyrosine.

In the initial step of GA synthesis, alanine, tyrosine, γ-O-benzyl glutamic acid, and trifluoroacetyl lysine are converted to reactive N-carboxy anhydrides (NCAs) by treatment with triphosgene. The benzyl and trifluoroacetyl protecting groups are employed to prevent cross-linking of the polymer chains. Next, the NCAs are dissolved in dioxane, after which the polymerization reaction is initiated by the addition of diethylamine (DEA) and allowed to proceed in a highly controlled manner. The resulting crude protected long-chain polymer is then isolated (Intermediate 1). In the subsequent step, Intermediate one is treated with hydrogen bromide in acetic acid under defined time and temperature conditions until the characteristic molecular weight distribution of GA is achieved and the benzyl protecting groups are removed from the glutamic acid side chain. This yields a second isolated intermediate (Intermediate 2), which in a subsequent step is treated with aqueous piperidine to remove the trifluoracetyl protecting groups and purified by diafiltration. Finally, resultant polymer is lyophilized to yield the final drug substance.

An extensive battery of tests was employed to demonstrate sameness of MGA to COP. Through the course of the development program, numerous individual and orthogonal tests were conducted probing different aspects of the products, including both chemical and biological tests. Here we report on the key tests that followed the FDA’s product-specific guidance on GA (i.e., physicochemical properties, structural signatures for polymerization and depolymerization, and results in a biological assay).

For quantitative analytical methods, equivalence ranges were established based on measured results for COP. For qualitative tests, visual comparisons of spectra or chromatograms were performed. In each of the quantitative analytical tests, a minimum of 20 lots of COP were used to define equivalence ranges. A minimum of three lots of MGA were analyzed in each test and each lot was evaluated individually against the equivalence ranges. For the EAE studies, the MGA lots used were also part of the analytical characterization program. For each EAE study at least 3 separate lots of MGA were included.

A brief description of the analytical methods is as follows. Amino acid analysis was performed using a well-established acid hydrolysis and chemical derivatization methodology. Molecular weight distributions were determined by size-exclusion chromatography. Total DEA was determined by gas chromatography following exhaustive hydrolysis of the polymer, while DEA-adducted amino acids were determined by tandem mass spectrometry. Circular dichroism (CD) analysis was performed by direct analysis on a spectrometer. Terminal sequencing was performed using Edman (N-terminal) or thiocyanate (C-terminal) chemistry with the resultant amino acid derivatives quantified by high-performance liquid chromatography.

The sensitivity of the analytical tests to detect potential differences in glatiramer products was assessed by means of various “negative control” polymers. Because the structural signatures in GA arise based on the fundamental chemistry of the synthetic process, intentional targeted alterations of the process provide a useful means to induce differences between otherwise similar polymers and therefore to test the discriminatory power of the methods.

Mylan’s characterization program included two “Stage 1” negative controls prepared by altering the polymerization reaction. One negative control (NC-1) was prepared using an alternate solvent in place of dioxane, while a second, NC-2, was made by altering the time relationship between monomer addition and initiation of the reaction. Additionally, a “Stage 2” negative control polymer (NC-3) was prepared by using an alternate reagent in place of HBr/HOAc for the partial depolymerization step of the process.

The intent when preparing the negative controls was to generate polymers that met the COP label claims for molecular weight and amino acid composition (US Food and Drug Administration, 2018d) (17), but that differed with respect to the targeted structural signatures. NC-1, -2, and -3 each met the aforementioned label claims (Mp between 5,000–9,000 Da; amino acid composition within the equivalence ranges shown in Figure 1).

Three different Experimental autoimmune encephalomyelitis (EAE) models were used to demonstrate the equivalence in vivo of 20 and 40 mg/ml lots of MGA and COP with regard to biological function. In mouse models, development of disease was induced by one of 1) immunization with myelin oligodendrocyte protein (MOG_35-55), 2) immunization with proteolipid protein (PLP_139-151), and 3) receipt of encephalitogenic T cells from donor mice immunized with PLP_139-151. All animal experiments performed in the manuscript were conducted in compliance with institutional guidelines.

The fundamental reaction scheme for the synthesis of GA entails the use “activated” forms (N-carboxyanhydrides) of the four constituent amino acids, polymerization in dioxane initiated by DEA, and partial depolymerization in HBr/HOAc (Campos-García et al., 2017). This underlying chemistry gives rise to various structural signatures in the resultant polymer due to, for example, the mechanism of initiation, the relative reactivities of the monomers, and the selectivity of peptide bond scission. MGA is manufactured using this same fundamental reaction scheme and therefore satisfies the first FDA criterion for GA active ingredient sameness.

Criterion two requires that the physicochemical properties of generic GA, including amino acid composition, molecular weight distribution, and spectroscopic fingerprints, are equivalent to COP.

Amino acid composition is one of the defining characteristics of COP, with the relative amounts of each amino acid listed in the product label ( Copaxone, 2021 ). As part of the sameness assessment, the amino acid compositions of MGA and COP were determined. Results of this analysis, expressed as mole fractions of each amino acid, are presented in Figure 1, and show that all MGA lots fell within the equivalence range.

As with amino acid composition, molecular weight is a defining characteristic of COP with an average value of 5,000–9,000 Da listed in the product label ( Copaxone, 2021 ). In addition to a simple “average” molecular weight analysis, the sameness assessment entailed a full characterization of the molecular weight distributions of the products, an important measure given that GA is a polydisperse polymer. Using size-exclusion chromatography (SEC), the statistical moments of number average, weight average, and Z-average were determined. Additionally, polydispersity index and the molecular weight corresponding to the SEC peak apex (believed to be the molecular weight measure specified in the COP label) were evaluated. Results are provided in Figure 2 and show MGA to be equivalent to COP across each measure.

FDA’s physicochemical properties criterion includes a requirement to assess “spectroscopic fingerprints” of the products. The key spectroscopic test used in the characterization of MGA was CD, which was selected based on its sensitivity to detect differences. CD is an optical spectroscopic technique that measures the differential absorption of circularly polarized light in a sample and is commonly used to characterize the secondary structure of proteins (Ranjbar and Gill, 2009). CD spectra obtained for three lots each of MGA and COP are shown in Figure 3. Similar spectral features (e.g., a positive peak around 190 nm and negative peaks around 208 and 222 nm) are seen for all lots of product.

Key structural signatures for polymerization include total DEA and the relative proportions of DEA-adducted amino acids. A well-known initiation mechanism of amine-initiated NCA polymerizations is the normal amine mechanism or (NAM) (Shunmugam, 2017), in which the amine acts as a nucleophile and attacks carbon 5 of the NCA, resulting in ring opening and the generation a new free amino (from the opened NCA) that attacks a second, intact NCA and thereby propagates the reaction. Mechanistic studies on GA polymerization have shown that the reaction proceeds through a NAM pathway (data not shown). One key consequence of this mechanism is that a DEA molecule is incorporated into each polymer chain. Another aspect of this mechanism is that the length of the initial polymer chains (crude, protected polymer at the Intermediate one stage) is affected by the concentration of DEA used in the reaction (higher DEA concentration = shorter chains). Different chain lengths of intermediate polymer can in turn alter the relative incorporation of amino acids into the chains during the propagation step of the reaction, and in turn alter local sequences. Therefore, the total amount of DEA incorporated into the final polymer, a direct reflection of the concentration used in the polymerization step, is a key process signature.

When DEA molecules covalently attach to the NCAs, and ultimately become incorporated into the polymer itself, they link to NCAs of all four amino acids. However, the relative proportions of DEA linkages to these amino acids are nether uniform nor random. Instead, the proportions are determined by the relative rates of reaction of DEA with each NCA, which in turn are a function of the relative concentrations of each NCA and the inherent reactivity (kinetics) of reaction (i.e., some NCAs react faster than others, all other factors equal). As such, the relative proportions of DEA-adducted amino acids serve as another important process signature with which to evaluate MGA and COP.

Total DEA determined for MGA and COP is shown in Figure 4A. As can be seen, all MGA lots fell within the equivalence range. DEA-adducted amino acids are presented in Figure 4B and also show equivalence between MGA and COP.

The observed relative proportions of DEA-adducted amino acids reflect the relative reactivities of the monomers, as discussed above, and illustrate the nonrandom aspect of GA synthesis. For example, the mole fraction of alanine-DEA in COP is approximately 0.65, whereas that of tyrosine-DEA is approximately 0.04. These differ markedly from the relative amounts of these amino acids in the bulk COP polymer (approximately 0.43 and 0.09, respectively, as determined from amino acid analysis). These results can be explained by the higher (faster) reactivity of alanine with DEA compared with tyrosine.

Because DEA-related attributes are reflective of the polymerization step of the process, it is not surprising that negative controls NC-1 and NC-2, in which the polymerization process was altered, failed equivalence criteria for at least one DEA test. As shown in Table 1, NC-1 failed for total DEA content and NC-2 failed for DEA-adducted amino acids.

Amino acid composition at the N- and C-termini was employed as probes for process-related signatures arising from the partial depolymerization step (e.g., preferential sites of peptide bond cleavage), as well as for “propagational shift.” Propagational shift refers to the differential monomer composition along a polymer chain, which in the case of GA, arises due to the differences in rates of amino acid incorporations (due to reaction rate differences similar to that discussed above for DEA). Terminal amino acid composition provides information on propagational shift because the termini that exist in the final product reflect amino acids initially in the interior (both C- and N-termini) as well as on the N-termini of the initial crude protected polymer.

Cycle 1 N- and C-terminal amino acid compositions for MGA and COP are shown in Figures 5A, B. In all cases, MGA fell within the equivalence ranges for both terminal sequence analyses.

As was the case with the DEA-adducted amino acids discussed above, the repeatable patterns observed in the relative proportions of amino acids at both the N- and C-termini reflect the nonrandom aspect of GA synthesis and imprinting of process signatures by the underlying chemistry. The prevalence of alanine in excess of its fraction in the bulk polymers (>0.5 vs 0.43) reflects a preference at this amino acid for peptide bond cleavage. In contrast, the abundance of C-terminal glutamic acid is depleted relative to the bulk (≈0.05 vs 0.14) and thus suggests a strong bias away from scission that creates C-terminal glutamic acid residues.

Furthermore, the negative controls were used to demonstrate that terminal amino acid sequencing is sensitive to alterations in propagational shift (polymerization step) as well as depolymerization selectivity. NC-1 and NC-2 were manufactured by altering the polymerization step; however, the depolymerization step was conducted using the standard HBr/HOAc reagent. Tellingly, both of these polymers failed the N-terminal sequencing test and NC-2 failed both the N- and C-terminal tests. This can be explained by alterations in relative incorporation of amino acids that translated into terminal amino acid compositions.

In contrast, polymer NC-3 was made using crude protected polymer (Intermediate 1) manufactured from the standard process and thus had no alteration in propagational shift. The change in the manufacture was in the partial depolymerization step, in which an alternate reagent was used. As seen in Table 1, NC-3 failed the C-terminal sequence test.

These data provide strong evidence that MGA and COP are equivalent in terms of structural signatures for polymerization and partial depolymerization.

Biological equivalence was determined by analysis of the impact of COP and MGA in three models of EAE–two preventive models and one therapeutic model. In all models of EAE, MGA and COP demonstrated similar efficacy with respect to their ability to block the induction of disease (Figures 6, 7, 8).

In the MOG-EAE induction experiment, treatment with either MGA or COP at 500 μg/mouse substantially blocked EAE induction in mice immunized with MOG_35-55. Incidence of disease was 100% in control animals and ranged from 17 to 42% in animals treated with 20 mg/ml lots of MGA or COP (Figure 6A, upper panel). Additionally, incidence of disease was 92% in control animals and ranged from 1 to 25% in animals treated with 40 mg/ml lots of MGA or COP (Figure 6B, lower panel). Both COP and MGA significantly and comparably reduced EAE incidence, postponed EAE onset, reduced mean maximum score, and reduced average end score compared with the vehicle group. In addition, all groups treated with MGA or COP had significantly higher average body weight than the vehicle group, which exhibited a decrease in body weight that paralleled disease onset and progression (Figures 9A, B, upper and lower panels, respectively). These analyses revealed no significant difference between the response to any MGA lot and the COP group response.

Similar results were obtained in the PLP_139-151-induced EAE model. Incidence of disease was 100% in control animals. Treatment with 20 or 40 mg/ml MGA or COP lots reduced the severity of the disease relative to the control group (Figures 7A, B , upper and lower panels, respectively). A slight decrease in body weight was noted in the control animal group after dosing and appeared to parallel disease onset and progression (Figures 10A, B, upper and lower panels, respectively).

In the adoptive transfer experiment designed to test the therapeutic effect of test agents after priming of encephalitogenic T cells, MGA and COP suppressed disease comparably. All the tested 20 mg/ml MGA and COP lots postponed EAE onset compared with the vehicle control group, with a median day of onset ranging from Day 13 to Day 15 for treated animals as compared with Day 9 for the vehicle control group (Figure 8A, upper panel). Treatment with 40 mg/ml MGA also suppressed EAE comparably to COP. All the tested MGA and COP lots postponed EAE onset compared with the vehicle control group, with a median day of onset ranging from Day 14 to Day 15.5 as compared with Day 10 for the vehicle control group (Figure 8B , lower panel). As anticipated with this model, mice within all groups exhibited weight loss coinciding with the development and progression of disease. (Figures 11A, B, upper and lower panels, respectively).

The economic burden of MS treatment and management is extensive involving various direct and indirect costs, medical and non-medical costs, and informal care costs. The cost of managing MS increases with patients’ age and disease progression. Studies have shown that among the various chronic diseases, MS is the second most cost intensive condition with respect to the direct medical costs incurred. Economic burden includes various costs such as diagnostics costs, physical and clinical examinations, interventions and monitoring, resulting in loss of productivity, employment and quality of life (QoL). Another major burden is the high cost of the disease-modifying therapies (DMTs). (Owens GM et al., 2016) (Garcia Dominguez JM et al., 2019)

Healthcare resource utilization (HCRU) is higher in MS patients compared with patient suffering from other chronic diseases. HCRU such as hospitalization, emergency department visits or frequent visits for physical, speech, or occupational therapy are higher in patients with de novo disease. (Owens GM et al., 2016). In 2010, the total lifetime cost per patient for the treatment of MS was estimated to be $4.1 million, of which, 75% of the total HCRU cost was towards treatment with DMTs. Therefore, a careful and delicate balance is required between a robust disease management and value generation when using high-cost DMTs. (Owens GM et al., 2016) (Garcia Dominguez JM et al., 2019)

Treatment and management of MS should be patient-centric and customized as per individual patient’s need. All the resources should be used cautiously. There should be greater collaboration between patients, healthcare providers, managed care organizations and health insurance carriers because many of the available drug therapies are considered therapeutically equivalent by payer pharmacy and therapeutics committees and an individual drug therapy may be determined by pricing contract. It should have plans to improve the clinical outcomes and QoL for patients with MS and thereby minimize the economic burden associated with complex chronic diseases. (Owens GM et al., 2016).