Conclusions
Residual HCP control and their risk assessment has been an industry-wide
challenge for biotechnology companies. This is mostly due to the
complexity and heterogeneity of HCP makeup in the upstream and
downstream process and the relative low abundance of HCP in final drug
substance. The latter makes analyzing residual HCPs like finding needles
in a haystack and thus requires highly sensitive and specific detection
methods that can detect and distinguish HCPs from a dominant matrix of
therapeutic proteins. A sandwich ELISA utilizing polyclonal antibodies
that can recognize and capture residual HCPs is often used as the
workhorse for residual HCP measurement. However, since the ELISA relies
on the antibodies to detect and quantify residual HCP amount in samples,
the ability of such antibodies to detect potential HCPs that can reside
in the process intermediates and final drug substance need to be
demonstrated and the method validated for its precision, accuracy,
linearity, range, LOQ and robustness (Gunawan et al., 2018). In
bioprocess development, a generic or platform HCP ELISA that is
commercially available or developed in-house is often used in the early
phase development up to the stage of process validation with appropriate
assay qualification to gain insight on the HCP clearance trend and batch
consistency. From phase III and beyond, either a platform assay or
upstream process specific assay is preferred to mitigate the risk of
inadequate coverage of HCPs specific to the manufacturing process by a
generic HCP ELISA (USP Monograph Chapter 1132). However, even with a
well-validated process specific assay, chances are that not all residual
proteins are quantified accurately given the difficulty to achieve 100%
coverage and to find relevant standards to quantify residual HCPs in all
process intermediates. More commonly, an upstream process mock HCP
culture from a null cell line is used as the calibration standard for
the ELISA assay. This often leads to quantitation error when certain
HCPs are enriched during this process, especially when the amount of
HCPs present is in excess of the antibodies available to capture and
detect the HCPs (Zhu-Shimoni et al., 2014). As indicated in Figure 3,
the use of different calibration standards can lead to significantly
different measurement of protein values, despite being uncommon to see
such a large extent of difference with well-qualified assay standard.
Therefore, orthogonal methods are often needed to supplement the results
obtained from ELISA testing to evaluate the overall risk of residual
proteins while the ELISA method needs demonstrate its fitness for its
intended purpose in early phase development and fully validated at late
phase development to demonstrate its precision, accuracy, linearity,
range, sensitivity, specificity, and robustness.
Unlike biologics, residual HCPs in small molecule APIs often have
distinct biochemical properties and can be easily separated from API by
SEC-HPLC or Tangential flow filtration (TFF). However, the use of TFF
and column-based separation is not desired in small molecule process
development. Process chemists tend to use phase cut and crystallization
as the main means of isolation (Wells et al., 2012; Wells et al., 2016).
In the case of MK-1454 discussed in this manuscript, an initial
isolation of API using these traditional process chemistry techniques
achieved high API yield and purity comparable to chemical synthesis.
However, E. coli proteins along with the enzymes added to the
process were not completely removed or polished, leaving a large pool of
proteins present at trace levels in the Prep. Lab batch API. The amount
of residual E. coli proteins were estimated using the commercialE. coli HCP kit while the enzymes used in the reaction are not
reactive to the kit antibody (Figure 4b and 4c). Although efforts have
been made and some success has been achieved to use the total input
level of proteins (reactive or not) to estimate the residual amount of
proteins in the API, quantification by this approach have the risk of
over-estimating E. coli proteins if contamination occurs during
the process. To overcome these challenges, 1D SDS-PAGE gel with silver
stain and LC-MS was used to estimate the total protein amount in API and
assess the risks associated with those proteins by their relative
abundance level and in silico predicted immunogenicity. Although the
proteins present in the Prep. Lab. batch API are not considered to pose
significant immunogenicity risk, these materials haven’t been assessed
for immunotoxicity in animal studies or clinical trials. Instead, the
chemically synthesized API was used in early clinical study and the
biocatalytic route is developed for commercial chemistry. To minimize
the potential immunogenicity risk and allow a direct use of biocatalytic
route synthesized API in clinical trials, further reduction of residual
proteins is achieved by process optimization. The new workup process has
barely any detectable level of proteins as analyzed by ELISA, 1D
SDS-PAGE with silver stain and proteomic LC-MS/MS. This case study
demonstrates the importance of a holistic analytical control strategy in
HCP characterization for biocatalytic route synthesized API. This
holistic analytical control strategy allows process chemistry to design
new commercial manufacturing process to remove residual proteins (HCP
and enzymes) to insignificant levels (<10 ng/mg) in three
representative batches of API. With a robust process and holistic
analytical characterization, a process-specific ELISA using antibody
reagents developed for matching cell lysate used for MK-1454
biocatalysis may not be needed in late phase development weighing in the
time and resources investment in developing such an assay, the API
comparability and the low demand in MK-1454 quantity in commercial
manufacturing. However, the holistic analytical characterization
presented here, together with the API stability monitoring, will be
essential to reduce patient safety and product quality risk associated
with the presence of residual E. coli HCPs and enzymes.