Fill–finish and aseptic biomanufacturing

Most sterile small-molecule products are sterilised in their final container: steam, dry heat or gamma radiation applied after sealing, with a target probability of a surviving organism of one in a million. A protein cannot take any of that. Heat unfolds it, radiation generates radicals that oxidise methionine and tryptophan and cross-link chains. So the sterility of a monoclonal antibody, a vaccine or a cell therapy is not achieved at the end — it is achieved by never letting an organism in. Every feature of a fill–finish suite follows from that inversion.

Sterility cannot be tested in

The compendial sterility test examines about twenty containers from a batch. If one container in a thousand were contaminated, that test would almost always pass. This is not a weakness to be fixed by testing more; sampling a rare event to a useful confidence would consume the batch. Assurance therefore has to come from the process, and the process is validated by aseptic process simulation — filling growth medium instead of product under worst-case conditions. EU GMP Annex 1 sets the expectation of zero contaminated units at a run size of at least 5,000, which supports a claim about contamination rate at a stated confidence, and it is that statistical statement, not the sterility test, that the release actually rests on.

The contamination control problem is aerodynamics

Airborne particles carry organisms, and the dominant source in a cleanroom is the human being in it — a gowned operator still sheds on the order of thousands of viable-carrying particles per minute. Grade A protection works by sweeping them away before they reach an open container: unidirectional airflow at roughly 0.36–0.54 m/s, with “first air” — air that has passed through the terminal filter and touched nothing else — reaching the critical surfaces. Grade A limits are 3,520 particles ≥0.5 µm per cubic metre and, in practice, effectively no recovered organisms. Since the operator is the source, the structural answer is to remove them from the airstream altogether: restricted access barrier systems, or closed isolators with vaporised hydrogen peroxide decontamination and glove-integrity testing. The 2022 revision of Annex 1, effective 2023, made this explicit by requiring a documented contamination control strategy rather than a checklist of rooms.

The product is also damaged by its container

Sterility is only one of the failure modes, and the others are physical chemistry. Silicone oil used to lubricate prefilled syringe barrels migrates into solution and its interface nucleates protein aggregation; tungsten residues left by needle-hole forming in glass can do the same. Type I borosilicate glass under certain formulation and processing conditions delaminates, shedding thin silicate lamellae. Pumps shear: rotary piston pumps are precise but abrade, peristaltic systems are gentler but less accurate. Lyophilisation sets its own trade — fast freezing gives small ice crystals and a large specific surface that can denature protein, slow freezing gives a cake that dries and reconstitutes better.

Finally the seal has to be proven, and dye ingress is a poor detector. USP <1207> reframed container closure integrity around deterministic physical methods — helium leak, vacuum decay, high-voltage leak detection — with a leak size, not a pass mark, as the meaningful quantity.