Drug substance versus drug product

Drug substance manufacturing ends at the frozen bulk; drug product manufacturing begins when that bulk is thawed, formulated, sterile filtered and filled into vials or syringes. The two stages sit in different dossier sections, 3.2.S and 3.2.P, are often run at different sites, and carry different GMP expectations.

ICH Q7 sets the GMP framework for active pharmaceutical ingredients, and its scope table applies increasing GMP stringency to biotechnology processes from maintenance of the working cell bank onward. Cell line construction and master cell bank preparation sit before that line; everything from vial thaw to bulk fill sits inside it, with batch records, qualified equipment, controlled materials and validated cleaning.

The specification that defines a successful batch follows ICH Q6B: identity, purity and impurities, potency, quantity and general tests such as pH and endotoxin. Drug substance manufacturing is designed backward from that specification: every unit operation delivers one or more attributes, and every in-process control confirms that it did.

Choosing the platform: mammalian versus microbial

The expression platform is set by the molecule, not by preference: proteins that need mammalian glycosylation or complex disulfide folding go to CHO or HEK293, while small non-glycosylated proteins go to Escherichia coli or Pichia pastoris. The choice fixes the shape of the entire drug substance manufacturing process.

Upstream GMP operations: seed train, production bioreactor, harvest

Upstream drug substance manufacturing has one objective: deliver a clarified harvest of known titer and quality to the purification suite on a defined day. Three operations achieve it.

Seed train. One working cell bank vial is thawed and expanded through shake flasks and seed bioreactors (N-2, N-1) until the inoculum reaches the production vessel's seeding density, commonly 0.3–0.5 × 10⁶ viable cells/mL for CHO. Each passage has a viability and density limit, and generation number is tracked against the in vitro cell age limit set with the end-of-production cell bank.

Production bioreactor. A fed-batch CHO culture runs 12–17 days with controlled temperature, pH, dissolved oxygen, agitation and feeds. Daily sampling measures viable cell density, viability, glucose, lactate, ammonia and titer. The harvest trigger is a viability floor (often 70–80%) or a fixed day, whichever comes first, because declining viability releases host cell protein and proteases that burden every downstream step.

Harvest and clarification. Cells and debris are removed by disc-stack centrifugation and depth filtration, or by depth filtration alone at single-use scale, then 0.2 µm filtration. Depth filter area is sized from turbidity and cell density data and confirmed in engineering runs. The unprocessed bulk is sampled here for the in vitro adventitious virus assay required by ICH Q5A(R2), plus mycoplasma and bioburden testing.

Downstream: capture, viral inactivation and clearance, polishing, UF/DF

Downstream drug substance manufacturing removes process- and product-related impurities in a fixed order and, for mammalian products, demonstrates orthogonal viral clearance. A typical antibody platform has five steps.

1. Capture. Protein A affinity chromatography binds the antibody from clarified harvest and elutes it at low pH into a concentrated pool. Loading is set by dynamic binding capacity at a defined residence time, and resin lifetime (often 100–200 cycles) is tracked by cycle count.
2. Low-pH viral inactivation. The capture eluate is held at pH 3.5–3.8 for a defined time, commonly 30–60 minutes, then neutralized. Hold pH, temperature and duration are critical process parameters because the log reduction value is validated only inside those ranges.
3. Polishing. One or two ion exchange steps remove host cell protein, DNA, aggregates and leached Protein A. Anion exchange in flow-through mode is the usual first polish; cation exchange in bind-elute mode resolves aggregates and charge variants. Hydrophobic interaction or mixed-mode resins substitute when the molecule demands it.
4. Virus filtration. A parvovirus-retentive filter provides size-based clearance that is mechanistically independent of low-pH inactivation. Filter area is sized from throughput studies at constant pressure.
5. Ultrafiltration/diafiltration. Tangential-flow filtration concentrates the pool and exchanges it into formulation buffer. Membrane loading, transmembrane pressure, crossflow rate and diavolume count (typically 7–10) are set at pilot scale and held through scale-up.

ICH Q5A(R2) expects clearance to be demonstrated across mechanistically different steps using model viruses, with a scaled-down model of each step justified against the production operation. For microbial products, the viral clearance block is replaced by endotoxin removal and, for inclusion-body products, refolding before capture.

In-process controls and release testing

In-process controls decide whether a batch proceeds; release testing decides whether it ships. Every drug substance manufacturing step has at least one control with a numerical limit.

Release testing follows the ICH Q6B specification: identity by peptide mapping or charge-variant profile; purity by SEC-HPLC, CE-SDS and ion exchange; impurities by host cell protein ELISA, host cell DNA qPCR and residual Protein A ELISA; potency by a cell-based or binding assay; safety by endotoxin and bioburden. Residual DNA limits commonly reference the 10 ng per dose figure cited in WHO guidance. A full release panel can take four to six weeks, so in-process data are used to pre-empt failures.

Bulk fill, storage and shipping

The final step of drug substance manufacturing is a closed transfer of formulated bulk into containers that will be frozen, stored and shipped to the drug product site without change in quality. Three decisions govern it.

Container and fill. Bulk is filtered through a 0.2 µm filter into single-use bags, bottles or carboys in a controlled environment. Fill volume is matched to the drug product batch size so that each thaw consumes whole containers. The filter integrity test and the fill-weight record are release documents.

Freezing and thawing. Uncontrolled freezing concentrates solutes and protein at the freeze front and can drive aggregation, so bulk is frozen in controlled-rate systems and thawed under the same validated program. Small-scale studies define the number of freeze/thaw cycles the product tolerates.

Storage and shipping. Frozen drug substance is typically stored at −40 °C or below, often at −70 °C, with a shelf life supported by real-time stability data under ICH Q5C. Shipping uses qualified containers with temperature loggers, and the receiving site verifies the temperature history before acceptance.