Table III: Illustrative "bottom-up" bioprocess QbD cost analysis for additional CMC workload on a per product basis. FTE is
full-time employee, DOE is design of experiments.
|
Table III shows the cost estimate for the additional CMC workload for an illustrative bioprocess, assumed to be executed using
internal company resources. Estimates have been made in areas of staff, expenses, and capital which add up to $6–10 million
over the course of a typical 6-year product CMC development cycle. Timeline impact from QbD-related workload was assumed neutral
because benefit of reduced time from less redoing of deliverables owing to CMC failures was felt to be offset by the extended
time needed for additional net effort increases. The value of $10 million was taken for the QbD implementation cost for this
analysis.
Even if the incremental costs associated with initial QbD workload are not as significant as perceived, they are paid during
drug development stages and the payback is largely postlicensure in manufacturing (27). Overall, QbD is a near-term investment
of staff, expenses, and some capital, funded by process and analytical development, for a long-term, postlicensure gain in
current operations or expected improvements that is realized by manufacturing (e.g., supporting knowledge for atypical investigations).
Because most companies have structurally distinct development and manufacturing organizations, it can be hard to orchestrate
this preinvestment across development stages and through to manufacturing. Thus, pre-investment must be carefully managed,
especially for higher risk pipeline products. More success may have been achieved to date solely by manufacturing applications,
for example the use of standardized work to reduce variability for licensed products (28). This success is reflected in the
productivity cost-avoidance benefits estimated in Table II.
Table IV: Published rough biopharmaceutical base numbers for "top-down" calculation of QbD implementation costs.
|
Initial high implementation cost estimates for the first few products in a company's process platform are expected to significantly
decrease as experience is gained for similar platform products, and quickly produce observable process development efficiencies.
Implementation costs are not expected to decrease, and may even increase, as the biopharmaceutical industry undertakes new
therapeutic modalities, often bringing in new platforms through partnering with or buying small biotechnology companies. Regardless,
because QbD is based heavily on design-for-six-sigma (DFSS) principles, there is still substantial net benefit in its application
even to novel aspects of a biopharmaceutical product or complex bioprocess steps where less prior knowledge is available to
be leveraged (29).
Table V: Biopharmaceutical QbD costs relative to total chemistry, manufacturing, and controls (CMC) costs and total development
costs. FTE is full-time employee.
|
QbD implementation costs, estimated from adding up the incremental costs for the additional CMC workload (see Table III) compared
reasonably well with the estimate obtained from published ranges for other typical biopharmaceutical industry costs for an
example of a product with $1 billion in annual sales (see Table IV). In both cases, the QbD implementation cost was 10–20%
of process/CMC development costs. Although the $10 million QbD implementation cost estimated for per-product QbD implementation
is a significant chunk (around 10–20%) of CMC development costs, it is a very small percentage (around 1–2%) of total product
development costs (see Table V).
Table VI: Illustrative business case return on investment (ROI) showing linear variation with product sales, discards arising
from insufficient understanding and batch cost.
|
With both potential QbD implementation costs and cost avoidance benefits being loosely quantified, a rough return-on-investment
can be calculated and thus the question of whether QbD is worth the investment can begin to be examined, albeit with some
uncertainty. Consequently, an illustrative business case was constructed to examine sensitivity of ROI to assumptions in product
sales, batch cost-of-goods, and the percentage of batches discarded owning to reasons of insufficient understanding (see Table
VI). The cost of QbD implementation was estimated at $10 million (see Table III). An overall discard rate (based on all reasons
for failure) of 5% and fraction of sales that is COGS of 25% (15% cost of goods only, plus 10% for cost of services) was used
to calculate potential benefit (compared with the more detailed estimates presented in Table II)(23). The simple benefit calculation
does not include reduced lost sales, reduced market share, regulatory implications, or other detriments associated with higher
discard rates/uncertain supply for a licensed product.
ROI varied linearly and directly with product sales, discards arising from insufficient understanding, and batch cost. Although
this set of calculations suggested attractive ROIs for processes with high discards, improving low discard processes also
had merit (see Table VI). Specifically, for a product with $1 billion in annual sales and $250 million COGs, a decrease in
discard rate from 5% to 1% translated to a $2.5 million annual savings opportunity, or a 4-year ROI.
|
