Scale-Up Dissolution Failure of a BCS Class II Weak Base Tablet
A 100 mg immediate-release tablet of a BCS class II weak base (pKa 6.4, logP 4.1, solubility 0.03 mg/mL at pH 6.8 and 9 mg/mL at pH 1.2) has been scaled from 20 kg to 300 kg. Composition: API 40%, microcrystalline cellulose 33%, lactose monohydrate 18%, croscarmellose sodium 5%, colloidal silicon dioxide 0.5%, magnesium stearate 1.0%. The 20 kg batch released 86% in 45 minutes. The 300 kg batch shows hardness 14–16 kp (target 8–12), friability 0.21%, disintegration 9 minutes, assay 99.1%, blend uniformity %RSD 1.8%. Dissolution in 900 mL pH 6.8 phosphate, 75 rpm paddle, gives 62%, 64% and 61% at 45 minutes against Q = 80%. In pH 1.2 medium the same tablets release 94% in 45 minutes; in pH 4.5 acetate, 76%. Laser diffraction of the API lot used gives d90 = 41 µm, against 18 µm for the previous lot. Magnesium stearate blending time was 10 minutes at 300 kg versus 3 minutes at 20 kg.
1. Construct the root-cause argument from these data, separating the contributions of particle size, lubrication and compression force.
Read the worked answer
Three variables moved together and must be separated.
- Particle size: d90 rose from 18 to 41 µm. For a BCS class II weak base with 0.03 mg/mL solubility at pH 6.8, dissolution rate is surface-area limited; a 2.3-fold increase in d90 roughly halves the specific surface area and is a plausible primary driver of 62% versus 86%.
- Lubrication: 10 minutes of blending with 1% magnesium stearate versus 3 minutes coats particles in a hydrophobic film, reducing wettability and blocking pores. This hits hardest exactly where dissolution is already solubility-limited.
- Compression force: hardness 14–16 kp against a target of 8–12 lowers porosity and increases tortuosity, slowing medium ingress.
The pH-dependence data are the diagnostic key: at pH 1.2 the API is ionised and highly soluble, so even the impaired surface cannot limit release (94%). At pH 6.8, where the base is largely un-ionised and poorly soluble, surface area and wettability become rate-limiting. Assay (99.1%) and blend uniformity (%RSD 1.8%) exclude content problems.
2. Design the smallest set of experiments needed to confirm the cause and set the corrective specification, including acceptance criteria.
Read the worked answer
Use one 300 kg blend split into sub-batches and run a 2³ factorial with centre points, n = 3 tablets per condition.
Factors: API d90 (18 vs 41 µm, achieved by milling), magnesium stearate blending time (3 vs 10 min), and hardness (9 vs 15 kp).
Responses: % released at 15, 30 and 45 minutes in pH 6.8; disintegration time; water uptake or contact angle as a wettability surrogate.
Acceptance criteria: 45-minute release at least 85%, no individual value below Q + 5%, and similarity to the 20 kg reference batch with f2 at least 50.
Expected outcome: particle size dominant, lubrication a significant main effect with a size interaction, compression force minor over 9–15 kp. If that holds, set the API specification at d90 of 20 µm or less, reduce the magnesium stearate blend to 3 minutes (with a lubricant sensitivity study to define the upper limit), and cap hardness at 12 kp. Confirm on 100 kg then 300 kg, and repeat dissolution at 3 months 40 °C/75% RH because a hydrophobic lubricant film can worsen on storage.
3. A marketing colleague asks whether the QC dissolution medium can simply be changed to pH 1.2, since the tablets 'pass' there. Respond with the scientific and regulatory reasoning.
Read the worked answer
No. Substituting pH 1.2 would make the product look acceptable while testing the least biopharmaceutically relevant condition.
For a base with pKa 6.4, gastric pH keeps the molecule ionised and freely soluble, but gastric residence is short and gastric fluid volume small. The rate-limiting step for absorption is dissolution in the near-neutral small intestine, where the drug is largely un-ionised and solubility is 0.03 mg/mL. That is exactly the condition the pH 6.8 method discriminates - and it is the condition in which the failing batch behaves differently from the batch that passed.
Changing the medium after seeing the result is method-shopping and would not survive a regulatory review; the discriminatory power of the method is a validation parameter that must be demonstrated, not adjusted to fit the data.
A biowaiver is also unavailable: BCS class II products with a narrow therapeutic index or with excipients that alter absorption do not qualify. Keep pH 6.8 as the release method, and if desired add pH 4.5 as a characterisation test alongside the pH 1.2 data.