Nicolas Fourrier, Director Biomarker and Biopharmaceutical Testing, SGS
Abhishek Saharia, Director of Marketing, DiscoveRx
Fiona Greer, Global Director, BioPharma Services Development – Life Sciences, SGS

The development of a biosimilar product requires extensive characterization at many stages. Initially, one needs to determine the exact structure and the variation of the originator molecule – i.e. establishing the Quality Target Product Profile (QTPP). Regulatory authorities throughout the world then require head-to head comparison at multiple levels to demonstrate that the proposed biosimilar is “similar or highly similar” to the reference product. In a step-wise regulatory pathway, structural and functional characterization has been promoted to center stage as it is essential to demonstrating biosimilarity prior to further clinical development. Indeed, decisions based on factors observed during initial analytical screening could result in a shortened clinical development process.

An integrated approach to characterization requires assessment of primary and higher order structure including post-translational modifications and variations, biological activities, product and process impurities and stability and degradation pathways. At the functional level, pharmacologic activity should be evaluated by appropriate in-vitro and/or in-vivo assays. In-vitro assays may include biological assays, binding assays and enzyme kinetics. Such assays can be used, in a comparative sense, to provide additional evidence that the activity and potency are similar, supporting data from structural characterization. Moreover, they may also be used to further explore the consequences of observed minor structural differences and explore structure-activity relationships.

However, these in vitro and in vivo potency assays can be expensive and time consuming with readouts that are often far downstream of the initial drug/target interaction. Cell-based assays with a readout proximal to the initial target are an attractive alternative. DiscoverX has developed a series of cell-based assays that rely on the native biology of target receptors to quantitatively measure drug potency. Their PathHunter enzyme fragment complementation (EFC) technology consists of the β-galactosidase (β-gal) enzyme split into two inactive components, the enzyme donor peptide ProLink (PK) and an enzyme acceptor (EA). When brought together in close proximity, PK complements with EA to form active β-gal. The active enzyme catalyzes the substrate to generate chemiluminescent light, which amplifies the signal to make for a highly sensitive assay (Figure 1).

The power of this EFC technology for potency analysis was illustrated by Lamerdin et al. using an assay developed for the angiogenesis inhibitor bevacizumab. Bevacizumab inhibits the growth of blood vessels in tumors by binding to vascular endothelial growth factor A (VEGF-A). This binding blocks VEGF-A activation of the kinase insert domain receptor (KDR), which is the primary VEGF receptor (KDR is also known as VEGFR2). When activated, KDR promotes proliferation of vascular endothelial cells. Bevacizumab potency has traditionally been measured by proliferation of primary human umbilical vein endothelial cell (HUVEC) growth.

HUVEC assays are challenging: the assays typically take 3-4 days and the cells are highly variable due to donor heterogeneity. There is also a loss of specificity due to the fact that cell proliferation can be induced by a number of other growth factors. In contrast, the PathHunter assay targets an early event in the KDR activation cascade: the VEGF-A induced homodimerization of the KDR receptor (Figure 1) and can be performed in less than 24 hours. The assay consists of HEK293 cells that have been bioengineered to contain the receptors tagged with the EA and PK molecules (Figure 1), and are responsive to activation through VEGF-A. The PathHunter Bevacizumab bioassay has also been developed into a kit that includes cryopreserved ready-to-assay cells, reagents necessary to run the assay and 96-well plates, too. The use of cryopreserved ready-to-assay cells further reduces assay time by eliminating the need to culture the cells prior to running the assay.

Figure 1: Pathhunter Bevacizumab Bioassay

Lamerdin and associates demonstrated bioequivalence between the PathHunter Bevacizumab Bioassay and HUVEC assays. Using the kit they found that VEGF-A induced dimerization of KDR had an EC50 consistent with that seen in HUVEC assays (1-6 ng/mL). They also found that a dose-dependent inhibition of dimerization by bevacizumab produced an IC50 of approximately 39 ng/mL, which is comparable the ED50 of 50 ng/mL produced by a traditional HUVEC proliferation assay.

The group did a number of studies to qualify the assay for potency studies. They tested samples of varying potency from 50–150% relative to a reference sample of 100% over 3 days. They found the assay to be accurate (95.9%) and precise (4.1%), with a relative standard deviation (RSD) of less than 6%. Plotting measured values against expected potencies, they arrived at an R2 value of 0.985. Finally, they tested three different lots and found <15% CV in EC50 between lots and 9.9–11% CV within lots.

At SGS, our Biomarker and Biopharmaceutical Testing laboratory has further qualified the PathHunter Bevacizumab Bioassay for use in bevacizumab and biosimilar potency testing. Initially we verified that VEGF165 activated the KDR at 12 different concentrations from 0.0483–200 ng/mL, including a negative control, with eight replicates each. The replicates were analysed as four sets of duplicates: lines (rows) A-B, C-D, E-F and G-H. The results demonstrated the expected 4-parameter dose response curve and the replicates had low CV% (Figure 2). The R2 was greater than 0.9 and the Top/Bottom (T/B) ration was greater than 3. The VEGF165 mean EC80 was 3.93 ng/mL.

Figure 2: Verification of KDR Activation by VEGF

Similarly we tested the neutralization of KDR induction with concentrations of bevacizumab ranging from 1.54–6000 ng/mL with a set concentration of VEGF165 at 3.93 ng/mL. Again, each concentration was examined in four sets of duplicates: lines A-B, C-D, E-F and G-H. Once again the results had a 4-parameter curve and the replicates had a low CV% (Figure 3). The R2 was greater than 0.9 and the T/B ratio was greater than 2. The mean bevacizumab IC50 was 31.2 ng/mL.

Figure 3: Verification of KDR Activation by VEGF

In analysing the data from a single bevacizumab dose response assay, we compared the results from all wells on the plate to the data without the outer row of wells on the plate. We also looked at the data using different sets of duplicates: A-E, B-F, C-G and D-H. Table 1 summarizes the results, which are all comparable at the level of mean IC50, but different at the level of imprecision (CV%). The assay results showed that the best performances in terms of precision were reached when duplicate of wells were farthest one from the other: A-E, B-F, C-G and D-H.

Table 2: Impact of Well Position on Bevacizumab Dose Response Assay

To qualify the assay we prepared two duplicate preparations of bevacizumab at 100% and 50% and tested them on three different plates and two different days. Table 2 presents the data for 100% bevacizumab tested on three different plates on the same day and Table 3 compares those same three plates with a plate (D) run on a different day. In all cases the data show good comparability with an inter-run mean IC50 of 45.6 ng/mL and an inter-day mean IC50 of 43.2 ng/mL.

Table 2: Comparison of Duplicate Preparations of 100% Bevacizumab in Three Separate Assays on the same Day

Table 3: Comparison of Duplicate Preparations of 100% Bevacizumab in Three Separate Assays on the same Day

Table 4 presents the results of the 50% bevacizumab assay. Again, there is good comparability between the plates and between the two days. The inter-day IC50 precision had a mean of 75.4 ng/mL with an SD of 18.8 and CV% of 25.

The accuracy of the assay was evaluated by comparing the IC50 obtained with bevacizumab solution at 50% strength to the IC50 obtained with bevacizumab solution at 100% strength. Relative accuracy was calculated using the following formulas:


The inter-run relative potency had a mean of 57% with an SD of 7 and CV% of 12. The inter-run relative accuracy had a mean of 115% with an SD of 14 and a CV% of 12. The IC50 had acceptable intermediate precision and good relative accuracy and precision at the level of relative potency. These data support the conclusion that this assay is appropriate for the evaluation of bevacizumab potency and its biosimiliars.

Table 3: Comparison of Duplicate Preparations of 100% Bevacizumab in Four Separate Assays on the Different Days

Because the development of a biosimilar product requires extensive characterization at the functional level, the PathHunter Bevacizumab Bioassay Kit provides an attractive alternative to more expensive and time-consuming assays such as the HUVEC assay. This class of assays are easy-to-use, commercially available, and highly reproducible. One additional benefit is the significant reduction in the time for assay development, translating into overall cost savings. The PathHunter assay has been shown to be equivalent to HUVEC assays for determination of activity and potency by Lamerdin et al. and has been qualified for use in our laboratory.


Lamerdin, J., H. Daino-Laizure, N.W. Charter, and A. Saharia Accelerating biologic and Biosimilar Drug Development 2016, BioProcess International, pp. 36-44.

ICH Q6B: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products

ICH Q2(R1): Validation of Analytical Procedures: Text and Methodology

USP guidance chapter “1032”: Development of Biological Assays

USP guidance chapter “1033”: Bioassay Validation

USP guidance chapter “1034”: Biological Assay Analysis