Regeneron’s High Profile Success? Demand for COVID-19 prophylactics – what role for Pichia expression technology to assist access

With Regeneron’s high profile, single-person, single-site, clinical trial proving successful (along with much supporting other clinical data!), will neutralising antibodies surf the second wave?

Our COVID-related blogs focus on capacity and cost of making the treatments and understanding opportunities or bottlenecks in large scale provision and new innovations. In this context, we review here the manufacturing costs at scale for nanobodies, and ask whether their production costs are really cheaper than CHO-produced mAbs.

An inhaled dose of COVID-19 neutralising nanobodies for prevention or early treatment of infection is the type of substantial development that can be catapulted to reality in times as these. A number of research groups are looking into the prophylactic treatment of COVID-19 using antibody-derived binding domains, or nanobodies (1-4, 5). The potential advantage of these smaller COVID-neutralising proteins (compared to mAbs), is their stability to nebulisation and delivery directly to airway tissues. AstraZeneca’s Flumist vaccine is the best example of nasal delivery for a biologic. However, for neutralising biologics the clinical proof-of-concept is sparse. Sanofi recently stopped development of an aerosolised prophylactic anti-RSV (ALX-0171), based on Ablynx’s nanobodies (6), after lack of efficacy in one Phase II trial (7). Indeed, a few of the current COVID-related developments refer to intravenous or subcutaneous dosing of the prophylactic first, before reaching for the nebuliser.

Independent of the application route of the nanobodies, their cost of manufacturing is claimed to be favourable compared to mAbs, potentially enabling global access. A production titre of 20g/L (1) for the molecules using a Pichia Pistorius expression system is at the heart of the claims.

In an earlier blog on COVID-19 biologics manufacturing capacity (based on mAbs), we planned for a short-term requirement of 2-3 metric tons of therapeutic with follow on capacity of 10 metric tons. The drug substance manufacturing cost was in the range of $50/g.

BioSolve Process has just introduced a new Pichia sp. process model, initiated, and verified by a leading large pharmaceutical company with Pichia speciality. We used the new model to analyse comparative production costs for either mAbs or nanobodies, at 15,000L bioreactor working volumes. Existing facilities are assumed to be utilised, so no new infrastructure modelled. The results are quite surprising at first, because despite the high throughput of the Pichia process – 10 days per batch compared to 34 days for CHO – the cost per gram is the same or slightly higher than CHO-produced mAb for the same expression titre. The reason is that at this large scale of production, the total consumables and bioreactor media costs represent around 55-80+% of the COGs, with labour around 10-20% and capital equipment contribution only 4-12%. The utilisation of consumables is directly related to the quantity produced, not how quickly that mass in produced. 10 metric tonnes of antibody can be produced with 2 bioreactors rather than CHO’s 6 bioreactors over the same time. However, the same quantity of adsorbents and filters are utilised, which dominate costs, particularly at higher titres (5-20g/L) when the cost contribution of the culture media is diminished. So once CHO cells are producing >5g/L mAbs, even a Pichia process will struggle to make significant savings under these conditions modelled. Including facility build investment or retrofit expenditure and evaluating different facility utilisation scenarios can all be achieved with the base Pichia (and mAb) models to plan for the most cost-effective process.

The good news is that at 5g/L titre, COGs (drug substance) for both mAbs and nanobodies is $19/g at the scale under question. At 10g/L the values are $13/g. Improving on the utilisation of the consumables with careful control of r costs, extending the working lifetime and processing capacity where possible, would be the next important factor in reducing costs further.