A drug substance, made by chemical synthesis or by a biological route. It is not measured in the same place, nor for the same decisions, as the finished product.
Making a drug substance means running a transformation: reaction, crystallisation, filtration, drying, or culture and purification. What is measured is therefore a degree of advancement, rarely a content.
The reference method is chromatographic or titrimetric, the decision window is the kinetics of the batch, and the attributes that decide (polymorphism, size, residual solvent, impurities) are not those of a tablet.
A probe sitting idle on a reactor? An immersion probe installed on a reactor or a crystalliser during development, and never taken on into production. Before buying anything new, the diagnosis: often the hardware is sound, and it is everything else that is missing. What to do with an instrument that has fallen out of use.
Where to start, depending on your role
- You run synthesis production: start from the step where the decision comes too late. Scoping.
- You develop the process: what a proof of concept has to settle, at pilot scale. Feasibility.
- You own quality: what software declared 21 CFR Part 11 covers, and what stays with you. 21 CFR Part 11.
- You decide the investment: the cost of a control, calculated on your batches. Cost of a control.
- You want examples: eight real situations, presented without names. Experience.
What changes, either side of the line
| Drug substance | Finished form | |
|---|---|---|
| What is measured | An advancement, a polymorphism, a size, a residual solvent | A content, a uniformity, a moisture, a thickness |
| The operation | Reaction, crystallisation, filtration, drying, isolation | Blending, granulation, compression, coating |
| The reference | Chromatography, titration, diffraction, gas chromatography | Chromatography, compendial test, loss on drying |
| The framework | ICH Q11 and ICH Q7, plus ICH Q6A on polymorphism | ICH Q2(R2), ICH Q14, USP <905> |
| Who you talk to | Chemical or process development, synthesis production | Formulation development, solid dosage production |
The right-hand column is the pharma page. The left-hand one is this page. A project that crosses both exists — that is scale transfer — and it is then scoped as two linked projects rather than one.
The operations where measurement pays, in chemical synthesis
The measurement physics is that of fine chemistry; what changes is the documentary constraint around it. The operations and technologies are detailed on the chemicals page, which describes the same reactors without the GMP layer.
- The reaction. Advancement, disappearance of the limiting reagent, appearance of an impurity. The end of reaction is observed rather than waited out, and over-reaction shows during, not at the final control. Following a reaction in-line.
- The crystallisation. The most rewarding operation of the segment and the least instrumented. Seeding, cooling profile, growth: the size distribution governs filterability, downstream drying and the specification. Measuring a size without diluting.
- Filtration and washing. End of wash, residual solvent, product carried into the mother liquor. A loss seen continuously instead of appearing in a mass balance.
- The drying. Residual solvent and moisture, with the same stake as elsewhere: stopping on the state reached rather than on a fixed duration. What a moisture measurement covers.
- Isolation and milling. Final particle size, and whether the polymorph survives milling. The point where an attribute won in crystallisation can be lost.
On the biological side
ICH Q11 covers drug substances that are chemical and biotechnological, and the line with the finished product is drawn the same way: culture and purification make the substance, formulation and aseptic filling make the product.
The operations, the sterile access routes and the two calibration regimes are treated on the biopharma page. What is worth keeping here is the symmetry: biological or chemical, what is measured upstream is a state of transformation, and the reference method is slower than the decision.
The framework is not the one of the finished product
Three texts govern this segment, and two of them appear nowhere in a solid dosage dossier.
- ICH Q11 — development and manufacture of drug substances. It defines what a critical synthesis step is and what has to be demonstrated about it. An in-line measurement finds its natural place there: it documents a step the final control no longer sees.
- ICH Q7 — good manufacturing practice for active substances. The regime differs from that of the finished product, and it begins at a step the dossier has to name.
- ICH Q6A — specifications, and in particular its decision trees on polymorphism: whether a polymorph has to be specified, and if so how it is controlled. That is the question which decides the measurement technology.
The rest of the framework is common to both worlds: what a multivariate procedure has to demonstrate, what a declaration of conformity covers, and the place of the measurement in the control strategy.
The two points particular to this segment
Polymorphism is not read by chance
Two polymorphs share a composition and differ in properties: solubility, stability, behaviour under compression. Near infrared answers sometimes, indirectly, through the lattice’s effect on hydrogen bonding and on scattering. RAMAN answers directly: its bands are narrow and attributable to a crystalline arrangement, which makes it the first choice as soon as the question is which form rather than how much.
The practical consequence is a selection criterion rather than a preference: a polymorph question settles the technology before any discussion of performance. What each family actually sees.
Residual solvent is decided when the dryer stops, not at the control
The reference method is gas chromatography, and it returns its result well after the dryer has been stopped. Running the drying to a fixed duration with a margin therefore means paying energy on every batch to cover the unfavourable one.
Following the solvent in-line moves the decision to the moment it has an effect. The reference method stays: it does not disappear, it becomes what the model is established against and then monitored on. What the reference campaign asks for.
What it has given elsewhere
Two of the published cases belong to this segment, and they are the two operations named above.
- A crystallisation known only afterwards — the size distribution observed at the end of the operation, with no way to correct the profile during it.
- A drying run to a fixed duration — the safety margin paid on every batch, for want of knowing where the solvent stood.
The cases, with what was done and what remained to be established.
Tell us at which synthesis step the decision comes too late. We will tell you whether it can be measured.
Forty-five minutes are enough: place the operation, say which measurement physics suits your medium, and name the reference method the model will have to be established against. If your question is about the finished product rather than the substance, we will tell you that too.