A version of this article previously appeared in August 2014 BioPharm International eBook supplement: Analytical and Bioanalytical Testing.

A number of biosimilars to insulin products are currently under development. Various modified insulins are available, which were engineered to have improved stability and duration of action over the natural molecule. As these branded insulin products start to reach the end of their patent life, there is a clear opportunity for competition. Several companies are already working on creating their own versions of these soon-to-be off-patent insulin products.

While it is fairly straightforward to prove that generic versions of small molecule drugs are equivalent to the original patented drug, this is not the case for biologics. The manufacturing process used to make a biologic has a real bearing on its precise structure and nature. This is why the term ‘biosimilar’ is used in place of ‘generic’ – the competitor product will not necessarily be identical to the original, but will be sufficiently similar in that it has the same therapeutic effect and side-effect profile.

No biosimilars have yet been approved by FDA, but in Europe the first, Sandoz’s version of the human growth hormone somatropin, was approved way back in 2006. There are now multiple versions of drugs such as human growth hormone, erythropoietin, follitropin-alfa and filgrastim. The first monoclonal antibody biosimilars, two versions of the anti-TNF product infliximab designed to treat autoimmune diseases, were approved in 2013.

An essential part of the development of a biosimilar is actually proving that similarity. The company must be able to demonstrate that its quality, safety and efficacy are comparable to the already-approved reference product via a panel of in vitro analytical tests. Once this similarity has been established, comparative preclinical and clinical studies that confirm biosimilarity in vivo must also be carried out. Only once the regulators are convinced that the two products do indeed have similar properties and effects in patients will they give it the marketing go-ahead.

The comparability of biological activity is determined via laboratory tests early on in the development process of a biosimilar. For insulin biosimilars, these will include potency bioassays that look at various mechanisms relevant to the insulin pathway in the body, including receptor binding and downstream mechanisms such as phosphorylation of the receptor.

The regulators put a great deal of emphasis on the results of these assays as they provide a good early read-out of how potent the biosimilar is going to be, compared to the reference product. Potential problems such as mutagenesis must also be investigated. Regulators want as many mechanisms as possible to be studied in vitro as this will increase the accuracy of predictions of what will be seen in vivo in those all-important clinical comparison studies later on. Two examples of assays that have been designed for the development of insulin biosimilars are a receptor binding competitive radioactive assay, and a mitogenesis potency assay.

Receptor binding assay

One of the most important things to establish during these bioanalytical studies is that the reference product and the potential biosimilar have comparability in reactivity and, if this comparability cannot be established, the likely reasons why not. This must be done using receptor binding studies or a cell-based assay.

Such a receptor binding assay has been developed by our laboratory, building on existing assay techniques.1,2 The binding affinity of insulin was measured using a competitive insulin receptor binding assay in which a recombinant soluble insulin receptor is incubated with insulin that has been labelled with 125-iodine and increasing concentrations of insulin. The radiolabelled insulin is competitively displaced by unlabelled insulin. Once the system reaches equilibrium, polyethylene glycol is added, and the labelled insulin and receptor complex precipitated via centrifugation. It is then washed, and the amount of radioactivity in the precipitate is measured using a gamma counter.

These counts per minute (CPM) read-outs are then plotted against the logarithm of insulin concentrations to a four-parameter logistic function to create dose-response curves. The IC50 is determined as the concentration of unlabelled insulin that is required to give 50% inhibition of labelled insulin binding.

Different concentrations of the insulin receptor were tested to determine the optimum concentrations for measuring binding while avoiding ligand depletion. The best ratio of specific signal over noise was obtained with 50ng/ml of insulin receptor, as shown in Figure 1. At this concentration, the ligand depletion was found to be negligible, as the percentage of bound tracer is below 15% as was previously demonstrated.3 The tracer concentration was set at 100 picomoles per liter, a concentration that leads to 30–60% of maximum signal, a level that is near to the equilibrium dissociation constant.

ar1 fig1

Next, the optimum incubation time also had to be determined. This should be sufficiently long enough to enable the ligand binding to reach equilibrium; literature reports suggested that the incubation should be carried out for 42 hours.1 Tests on the new assay were performed to see how a shorter incubation time of 18 hours performed compared to the results from a 42 hour incubation. There was no significant difference between the two, as shown in Figure 2, so the shorter incubation was deemed to be sufficient for a successful assay in this case.

ar1 fig2

Additionally, the optimum concentration range for unlabelled insulin was also studied. The entire dose–response curve was established over 13 concentration points ranging from 1 to 100,000 picomol/liter being tested, as shown in Figure 3. This established the optimum dilution range.

ar1 fig 3

Once the optimum conditions for the test had been determined, its effectiveness in establishing the comparability between two forms of insulin was assessed in a validation study. The behaviours of European Standard insulin EDQM and Eli Lilly’s Humulin R in the assay were measured and compared during the validation study to test the performance of the method in determining the relative potency using IC50 values from the two forms of insulin. Data from this validation study showed an inter-run precision on relative potency of <24%. The relative accuracy, or dilution linearity, of Humulin R at 50% and 200 % compared to Humulin R 100% were 98.54% and 102.95%, respectively. The robustness of the method was demonstrated by comparing data generated from runs performed by two different analysts. In addition, the validation demonstrated the stability of the critical reagents.

The results of this validation study confirm that this insulin receptor binding assay can be used to evaluate the binding affinity of insulin products. It is therefore suitable as one of those essential in vitro comparability studies required ahead of a clinical programme in the development of biosimilar insulin.

Mitogenesis potency

In the body, insulin binds to the native insulin receptor, but also the insulin growth factor-1 receptor, or IGF-1R. In normal physiological conditions in vivo with low circulating levels of insulin, insulin has a much higher affinity for its own receptor, and does not bind to IGF-1R. At higher concentrations, when it does bind to IGF-1R, mitogenesis is initiated. This would be a problem in patients, as the result is cell proliferation. It is therefore important to show that the binding affinity to IGF-1R of the reference product and the biosimilar are comparable if these side-effects are to be avoided. The safety work has already been done for the reference product, so this is not a safety study – rather, it is an assay to prove that the safety studies that were carried out ahead of the reference product’s initial registration will also be applicable to the biosimilar.

A new assay has been developed by SGS to assess comparative mitogenesis potency in insulin biosimilars. Initial development experiments compared two different cell lines, the osteosarcoma cell line Saos-2 and the breast cancer cell line MCF-7, as both of these cell lines express IGF-1 receptor at high levels. The cells were cultured using published optimized conditions,1,3 and both European Standard insulin EDQM and Humulin R were used to induce in vitro proliferation in the two cell lines.

Two colorimetric methods were used to quantify the proliferative response in side-by-side comparison experiments on the two cell lines: MTT, which measures mitochondrial activity, and BrdU, which quantifies the amount of DNA present. The MTT method showed no response in either cell line. However, a greater dose response was observed with the MCF-7 cells than for Saos-2 cells when the BrdU method was used.

Cell density was also optimized and the optimum cell seeding level for the assay determined to be 8000 cells per well. These were cultured in EMEM medium containing 10% FBS and 0.01mg/ml of bovine insulin for 18–24 hours prior to a 24 hour starvation of the cells using serum-free EMEM medium with 0.5% added BSA. They were then treated with standard insulin or Humulin R for a further 24 hours before data analysis.

Furthermore, the cell passage effect was also investigated, and data showed that cell passage two is more suitable (Figure 4) in this assay because the variability of the EC50 value was tighter and the assay length was shorter.

ar1 fig4

Optimum dose-response curves were then established for the MCF-7 cell line in the proliferation assay using Insulin EDQM and Humulin R. A range of dilutions from each sample were run to evaluate potency testing potential; as an example, the curve for mitogenic response of MCF-7 cells to dilutions of Humulin R is shown in Figure 5. The EC50 value was determined as the half-maximal effective concentration; in other words, this is the concentration of insulin that induces a response half-way between the baseline and the maximum response. The relative potency of the standard insulin and Humulin R were compared, and the relative potency calculated as the ratio of the EC50 values from the Insulin EDQM and the Humulin R (Figure 6).

ar1 fig5

ar1 fig6

The validation results showed an inter- and intra-run precision of the EC50 values, and the relative potency was lower than 30% (CV %). The dilution linearity was established for both 50% and 200% strength preparations of Humulin R at 111 % and 108 %, respectively. Finally, the pre-defined system suitability and sample acceptance criteria (Table 1) were applied and verified during this validation study. These will be used for run acceptance in in vitro comparability studies using this method.

ar1 fig7

Ultimately, the MCF-7 cells in combination with BrdU was selected as a suitable cellular model for the in vitro comparability of the mitogenic response of this cell line to insulin RMP and biosimilar compounds, and was validated according to USP <1033> guidelines.4 These, essentially, evaluate reproducibility between different analysts and between runs, and the relative accuracy, which can also be defined as assay range or dilutional linearity. The design of the bioanalytical run used during validation reflects conditions that are suitable for establishing the comparability of biosimilars in terms of mitogenesis potency.

With the likelihood of biosimilar competition for modified insulins within the foreseeable future, it is essential that the testing protocols are in place ahead of patent expiry if those competitor products are to be launched as soon as the patents expire. These two tests represent an important part of that strategy.


1 P. Kurtzhalset al.Diabetes 2000, 49, 999
2 L. Schäfferet al. PNAS 2003, 100, 4435
3 J. Chappell et al. J. Biol. Chem. 2001, 276, 38023-8
4 USP, <1033> Biological Assay Validation, 2010.


Rabia Hidi
Managing Director – Biomarkers & Biopharmaceutical Testing
SGS Life Science Services

Nicolas Fourrier
Team Leader – Biomarkers & Biopharmaceutical Testing
SGS Life Science Services

Catherine Diot
Team Leader – Biomarkers & Biopharmaceutical Testing
SGS Life Science Services

Alain Renoux
Vice President Bioanalysis – Laboratory Services
SGS Life Science Services