--- review: 2023-moseson-taylor-crystallinity-complex-cqa-asd title: "Crystallinity: A Complex Critical Quality Attribute of Amorphous Solid Dispersions" authors: "Dana E. Moseson, Lynne S. Taylor" year: 2023 journal: "Molecular Pharmaceutics" doi: "10.1021/acs.molpharmaceut.3c00526" pages: 24 --- # Crystallinity: A Complex Critical Quality Attribute of Amorphous Solid Dispersions **Scope.** The Perspective (Mol. Pharmaceutics 20, 4802–4825; received June 2023, published September 2023) treats crystallinity in amorphous solid dispersions (ASDs) as a critical quality attribute. Its organising idea is that crystals reach an ASD by two pathways, incomplete amorphization during processing and crystal creation (nucleation and growth) during processing or storage, and that their effect on release depends on the crystal origin and properties, the matrix, the supersaturation of the test and the drug. It reviews detection methods, non-sink release testing and the sink index, reference crystal materials, seven case studies (indomethacin and bicalutamide extrudates, naproxen, a Kinetisol drug, fenofibrate tablets, tacrolimus capsules, miscible and immiscible PVP dispersions) and closes with a risk-assessment pyramid and a caution against fixed crystallinity specifications. **Summary of the check.** The wiki matches the review closely on its own research line (Moseson and Taylor 2018–2022 on the critical temperature, residual crystals, defect-driven fragmentation, matrix crystallization of bicalutamide) and on tacrolimus, and in many places adds later or more detailed evidence (posaconazole–HPMCAS residual crystals by SHG, Shah 2024; paclitaxel seeds, Que 2018; delamanid and praziquantel two-stage tests, 2023–2025; six-bile-salt media). Three real differences remain: the review's case-study reading that indomethacin seeds "do not grow" against the wiki's record of extra desupersaturation by residual crystals from the same paper; the review's claim that SHG quantifies crystallinity in the parts-per-billion range against a lowest quoted SHG limit of about 10 ppm (calculated) in the wiki; and the review's hedged suggestion of subambient storage against the wiki's nilutamide evidence that nucleation can be fastest far below Tg. The review's printed classical nucleation equation lacks the minus sign in the exponent. The main gaps are regulatory specification (ICH Q6A, QC dissolution limits), Kinetisol and mechanochemical routes, 19F ssNMR and micro-CT, the naproxen and fenofibrate case studies, humidity-shifted PC-SAFT phase diagrams, and downstream drug-product processing. **Counts.** agrees 16 · wiki narrower 9 · wiki extends 23 · differs 4 · wiki silent 9 ## 1–2. Introduction and the materials science tetrahedron ### R1. Solid-state form of an ASD is a specified quality attribute under ICH Q6A - **Review** [p. 1]: more than 30 US FDA-approved products containing an ASD existed by 2023; ICH Q6A requires that a solid-state form (including the amorphous form) be specified, monitored and controlled when it can affect performance, bioavailability or stability, so specifications for crystalline content are sought, but a one-size-fits-all limit cannot be used. - **Kind**: cited (ICH Q6A; Bhujbal 2021) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki has no finding on regulatory guidance or product counts; the system dependence of the risk is covered (R59). ### R2. Homogeneity and chemical purity are CQAs beside complete amorphization - **Review** [p. 3]: besides full transformation of the crystalline drug, the ASD intermediate must be homogeneous, since lack of homogeneity from poor miscibility, high drug loading, poor mixing or phase separation may accelerate crystallization; in thermal processing drug and/or polymer may degrade chemically. - **Kind**: framework - **Wiki**: [Solid state stability F8](../../solid-state-stability/findings.md#f8): "A drug-rich amorphous phase decides crystallization, whether it is present from casting, made by sorbed water or set by the drug loading"; [ASD process F35](../../asd-process/asd-process/findings.md#f35): "a lower limit set by residual crystals and an upper limit set by chemical degradation". - **Verdict**: wiki extends - **Note**: The wiki supports both points with data: separated dispersions crystallized within 22 h to 6 days against 46 and 125 days for mixed ones (Rumondor 2009), and the bicalutamide–PVPVA extrusion that removed crystals degraded about 5% of the drug (Moseson 2020, 2021). ### R3. Crystallinity can lower bioavailability by three mechanisms - **Review** [p. 3]: crystals (1) do not contribute to the solubility advantage, (2) can initiate further crystallization, lowering dissolution rate and extent and causing desupersaturation, and (3) can grow during storage. - **Kind**: framework - **Wiki**: [residual crystallinity](../../concepts/residual-crystallinity.md): crystals that remain after processing "lose the amorphous solubility advantage, may accelerate crystallization on storage and can consume supersaturation in solution". - **Verdict**: agrees - **Note**: Same three mechanisms, taken in the wiki from Moseson 2018. ## 3. Formation pathways of crystallinity in ASDs ### R4. Crystallinity arises by incomplete amorphization or by crystal creation - **Review** [pp. 3, 2]: incomplete amorphization (crystal dissolution or size reduction not completed) or nucleation and crystal growth; the latter usually during long or accelerated storage but possible during manufacture (Figure 2). - **Kind**: framework - **Wiki**: [residual crystallinity](../../concepts/residual-crystallinity.md): "Small amounts of crystalline drug that remain in, or appear soon in, a nominally amorphous sample after preparation"; [Overview F6](../../overview/findings.md#f6): "extrusion below the solubility temperature leaves them". - **Verdict**: agrees - **Note**: The wiki covers both origins but does not name them as a two-pathway scheme. ### R5. ASD manufacturing methods fall into solvent-based, thermal and mechanochemical classes - **Review** [p. 3]: solvent-based processes (spray drying, coprecipitation, rotary evaporation, vacuum drying, freeze-drying, supercritical fluids) are more common because they suit many compounds and spare material; HME is solvent-free, continuous and scalable for thermally stable systems; mechanochemical activation (mostly cryomilling) has no commercial ASD product. - **Kind**: cited (Bhujbal 2021 and others) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki uses spray drying, rotary evaporation, HME and cryomilling as preparation routes in findings but has no comparison of the method classes. ### R6. Incomplete amorphization leaves residual crystals when temperature or mixing kinetics are insufficient - **Review** [p. 3]: relevant to HME, Kinetisol, microwave irradiation and milling; residual crystals remain if insufficient temperature or mixing kinetics (mechanical input, residence time) is provided; in solvent processes undissolved drug in the feed also leaves crystals. - **Kind**: framework, cited (Moseson 2018, 2022 and others) - **Wiki**: [ASD process F9](../../asd-process/asd-process/findings.md#f9): "Below the critical temperature no extrudate was crystal-free; at and above it residence time and input crystal size decided". - **Verdict**: wiki narrower - **Note**: The wiki's evidence is HME only (indomethacin, bicalutamide, posaconazole); Kinetisol, microwave, milling-induced amorphization and undissolved spray feed are not covered. ### R7. The formulation critical temperature is the lowest temperature at which all crystalline drug can dissolve in the molten polymer - **Review** [pp. 3–4]: Tc, also called the solubility temperature Ts, lies below the drug melting point for a miscible pair; DSC melting-point offset (approach from heating) and isothermal dissolution in an extruder (approach from undersaturation) are thermodynamically equivalent. - **Kind**: framework, cited (Moseson and Taylor 2018) - **Wiki**: [formulation critical temperature](../../concepts/formulation-critical-temperature.md): "the lowest temperature at which a homogeneous molecular dispersion of a given drug–polymer composition can in theory form, because all drug crystals can dissolve". - **Verdict**: agrees - **Note**: [ASD process F8](../../asd-process/asd-process/findings.md#f8) also records the equivalence argument of the 2018 paper and that Tc is not a critical solution temperature. ### R8. Flory–Huggins melting point depression (eq 3) and the sign of χ - **Review** [p. 4]: $\frac{1}{T_c}-\frac{1}{T_m}=-\frac{R}{\Delta H}\left[\ln\phi+\left(1-\frac{1}{m}\right)(1-\phi)+\chi(1-\phi)^2\right]$, with $T_m$ and $T_c$ in kelvin; depression arises from a favourable entropy of mixing and, for some systems, an exothermic mixing enthalpy; a negative or small positive χ indicates a miscible system. - **Kind**: framework (Flory–Huggins; Moseson and Taylor 2018) - **Wiki**: [ASD process F8](../../asd-process/asd-process/findings.md#f8) prints the same equation and records "PCZ–HPMCAS has a positive χ (+0.43) and lowered the melting offset by only about 7 °C". - **Verdict**: wiki extends - **Note**: Same equation. The wiki adds that χ depends on which point of the melting event is read (IDM–PVP K12: −6.16, −1.94 or −0.82; Marsac 2009) and gives a measured small-positive-χ case consistent with the review's criterion (Shah 2024). ### R9. Phase diagrams can also be built by recrystallization, PC-SAFT, empirical models and rheology - **Review** [p. 4]: besides DSC melting point depression, the recrystallization method, PC-SAFT and empirical models determine the solubility temperature, and rheology has recently been used as an alternative means of determining Tc. - **Kind**: cited (refs 53–60) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki's Tc values all come from DSC offsets extrapolated to zero heating rate. ### R10. Below the Tg line, compositions left of the solubility line are thermodynamically and kinetically stable, those right of it only kinetically stabilized, where most ASDs lie - **Review** [pp. 4, 16]: region 1 (above the solubility curve, below Tg) has thermodynamic and kinetic stability; region 2 (below the solubility curve, below Tg) only some kinetic stability; most ASD compositions at ambient conditions are likely in region 2. - **Kind**: framework - **Wiki**: [Miscibility F9](../../solid-state-stability/miscibility-and-drug-polymer-interactions/findings.md#f9): "the predicted solubility of the crystal in PVP is low, so most dispersions are supersaturated and their stability is inferred to be kinetic". - **Verdict**: agrees - **Note**: Same conclusion, reached in the wiki for PVP dispersions (Marsac 2006, 2009). ### R11. Amorphization by dissolution into the melt is kinetic: Noyes–Whitney and Stokes–Einstein - **Review** [pp. 3, 5]: $\frac{dC}{dt}=\frac{DA}{hV}(C_s-C)$ (eq 1) and $D=\frac{k_BT}{6\pi\eta r}$ (eq 2); dissolution is accelerated by high surface area, high diffusion coefficient, high temperature and intense mixing. - **Kind**: framework - **Wiki**: [Noyes–Whitney equation](../../formulas/noyes-whitney-equation.md): "$\frac{dM}{dt} = \frac{DA}{h}\left(C_s - C\right)$"; [Stokes–Einstein equation](../../formulas/stokes-einstein-equation.md): "The same expression, $D=k_BT/(6\pi\eta r)$, is printed for a solute in a polymer melt"; [ASD process F11](../../asd-process/asd-process/findings.md#f11): "Without mixing, drug crystals dissolve slowly in a polymer melt". - **Verdict**: wiki extends - **Note**: The review's eq 1 is the wiki form divided by the volume. The wiki adds hot-stage times (under 30 s at 161 °C to about 20 h at 121 °C for IDM–PVPVA) and a fitted melt dissolution rate constant of 0.1002 µm/s, 5–40 times below assumed literature values (Moseson 2018, 2022). ### R12. Crystal defects accelerate amorphization by fragmentation, confirmed by population balance modelling - **Review** [pp. 5–6]: TEM of residual crystals suggests mechanisms beyond diffusion; without mixing, defects increased the area for dissolution by fragmentation; in HME the same was "confirmed" by population balance modelling; using crystals of greater defect density can accelerate amorphization. - **Kind**: cited (Moseson 2018a, 2019, 2022) - **Wiki**: [ASD process F13](../../asd-process/asd-process/findings.md#f13): "Crystal defects are proposed as the starting points of channels and fragmentation ... but defects and dissolution rates were not measured"; [ASD process F37](../../asd-process/asd-process/findings.md#f37): "the fit uses the same data and breakage in the extruder was not observed". - **Verdict**: wiki narrower - **Note**: Fragmentation itself is observed in the wiki ([F12](../../asd-process/asd-process/findings.md#f12)), but the wiki records that defect density was inferred from SEM appearance and agglomerate share, so the review is more general than its evidence on the defect role and on "confirmation" by the model. ### R13. The 50 °C below Tg rule of thumb and its exceptions - **Review** [p. 6]: the common rule for a stability period of years is storage 50 °C below the sample's Tg, but several ASDs have reached years of physical stability under much less stringent conditions. - **Kind**: cited (Hancock 1995; refs 65, 66) - **Wiki**: [Crystallization tendency F4](../../solid-state-stability/crystallization-tendency/findings.md#f4): "storage about 50 °C below Tg has been proposed as a good way to prevent crystallization"; "Nucleation can be fastest far below Tg, so cold storage does not protect every drug glass". - **Verdict**: wiki extends - **Note**: Both treat the rule as approximate. The review gives exceptions in the lenient direction (ASDs stable closer to Tg); the wiki adds a neat-drug exception in the other direction (nilutamide nucleating most at −30 °C, about 60 °C below Tg; Trasi 2014). The wiki has no ASD stored for years near Tg. ### R14. Physical stability risk factors - **Review** [p. 6]: drug crystallization tendency, polymer selection and drug–polymer interactions, overall molecular mobility, drug loading above what the polymer can stabilize, and water-induced plasticization and/or phase separation; storage conditions and packaging matter during shelf life and after dispensing. - **Kind**: cited (refs 13, 14, 19, 64, 69–77) - **Wiki**: [Solid state stability F7](../../solid-state-stability/findings.md#f7): "a few percent removes most nucleation, fast growers need more, the drug loading must stay below the miscibility limit"; [Physical stability F12](../../solid-state-stability/physical-stability/findings.md#f12): "Sorbed water splits many initially mixed drug–PVP and drug–PVPVA dispersions ... into a drug-rich and a polymer-rich amorphous phase". - **Verdict**: wiki extends - **Note**: The wiki covers each factor with data and adds that bulk mobility measures do not rank crystallization rates ([Physical stability F5](../../solid-state-stability/physical-stability/findings.md#f5)); packaging is not treated. ### R15. Classical nucleation theory: rate expression and dependencies - **Review** [p. 6]: primary nucleation (no crystals, higher barrier) and secondary nucleation (crystals present, lower barrier); $J = A\exp\left[\frac{16\pi\gamma^3\upsilon^2}{3(k_BT)^3(\ln S)^2}\right]$ (eq 4, printed without a minus sign in the exponent); higher temperature and supersaturation raise nucleation rates; faster agitation in solution promotes nucleation; drug–additive interactions can hinder or promote self-assembly. - **Kind**: framework (CNT) - **Wiki**: [classical nucleation theory](../../formulas/classical-nucleation-theory.md): "$J = A\exp\left(-\frac{16\pi\gamma^{3}v^{2}}{3k^{3}T^{3}(\ln S)^{2}}\right)$"; the same page notes one source "(printed without the minus sign)". - **Verdict**: differs - **Note**: As printed, eq 4 would make J fall as S rises, contrary to the review's own text; the wiki's form with the negative exponent is the correct one. The review shares the typesetting error the wiki records for Elkhabaz 2021; the stated dependencies agree. ### R16. Supersaturation as a chemical-potential ratio, approximated by a concentration ratio - **Review** [p. 6]: $\ln S = \frac{\mu-\mu^*}{RT} = \ln\frac{a}{a^*} = \ln\frac{\gamma c}{\gamma^* c^*}$ (eq 5, the second equals sign missing in print); for dilute aqueous solutions without solubilizers $\gamma/\gamma^* \approx 1$ and $S = c/c^*$ (eq 6). - **Kind**: framework - **Wiki**: [activity-based supersaturation ratio](../../formulas/activity-based-supersaturation-ratio.md): "$\ln S = \frac{\mu - \mu^{*}}{RT} = \ln\frac{a}{a^{*}} = \ln\frac{\gamma c}{\gamma^{*} c^{*}}$"; [supersaturation ratio](../../formulas/supersaturation-ratio.md): "valid only if the activity-coefficient ratio γ/γ* is 1 (dilute solution, no solubilizer)". - **Verdict**: agrees - **Note**: Identical definitions and validity condition. ### R17. The extent, not the existence, of supersaturation decides crystallization; polymers widen the metastable zone - **Review** [p. 6]: a metastable zone often exists where nucleation is favoured but not observed in the time scale; low supersaturation can persist for long times; polymers can change the metastable zone width, and nucleation inhibitors let higher supersaturation persist longer. - **Kind**: framework, cited (refs 80–82) - **Wiki**: [metastable zone](../../concepts/metastable-zone.md): "The range of supersaturation in which a solution is supersaturated with respect to the crystal but does not nucleate spontaneously"; [Crystallization from solution F14](../../dissolution/d-crystallization-from-solution/findings.md#f14): "At lower supersaturation an inhibitor acts far longer". - **Verdict**: wiki extends - **Note**: The wiki quantifies the dependence (HPMC lengthened celecoxib's induction time 31-fold at S 7.3 against about 5-fold at S 14.7; Xie 2016) and shows that above the amorphous solubility droplets make inhibition harder ([D F37](../../dissolution/d-crystallization-from-solution/findings.md#f37)). ### R18. Crystal growth follows a power law in the driving force; polymers adsorb and slow growth, and growth can stop while supersaturation persists - **Review** [pp. 6–7]: $R_G = K_G(C_x - C_{eq})^g$ (eq 7) combines diffusion- and reaction-controlled growth; polymers reduce growth by adsorbing on the crystal surface; with inhibitors present growth may be arrested although supersaturation persists, so the concentration may not return to the crystalline solubility in a relevant time. - **Kind**: framework, cited (Mullin; Moseson 2020; refs 85–88) - **Wiki**: [power-law crystal growth rate](../../formulas/power-law-crystal-growth-rate.md): "$R_G = k_G\,A\,(C_b - C^{*})^{g}$"; [Supersaturation from ASD dissolution F22](../../dissolution/b-supersaturation-from-asd-dissolution/findings.md#f22): "HPMCAS ... could not stop growth on seeds ... a residual supersaturation far above the crystalline solubility remained". - **Verdict**: wiki extends - **Note**: The wiki adds the area term, measured growth orders (1.5 to about 2 for felodipine; 1.677 for tacrolimus), adsorption seen by AFM ([ASD process F25](../../asd-process/asd-process/findings.md#f25)) and the weakening of inhibition with rising supersaturation ([D F9](../../dissolution/d-crystallization-from-solution/findings.md#f9)). ### R19. In supercooled liquids and glasses molecular mobility is often the rate-limiting step - **Review** [p. 7]: the principles of solution crystallization apply to melt crystallization, where hindered mobility in a deeply supercooled liquid or glass prevents or substantially delays self-organization despite a favourable driving force. - **Kind**: framework, cited (reviews 19, 89–92) - **Wiki**: [Crystallization tendency F2](../../solid-state-stability/crystallization-tendency/findings.md#f2): "Crystal growth rate passes through a maximum at roughly 0.90–0.96 of the melting temperature"; [Physical stability F5](../../solid-state-stability/physical-stability/findings.md#f5): "Dispersion Tg, Tg-scaled temperature and bulk mobility measures do not rank polymers or drugs by crystallization rate". - **Verdict**: wiki narrower - **Note**: The growth maximum below Tm is consistent with mobility limiting at deep supercooling, but the wiki finds that bulk mobility measures do not rank drugs or polymers and that nucleation can be fastest below Tg, so the review's general statement is broader than the wiki's evidence supports for ranking purposes. ### R20. Processing itself can nucleate and grow crystals: solvents, moisture, drying, cooling and drug loading - **Review** [p. 7]: in cosolvent systems with dissimilar evaporation rates the drug may crystallize or phase separate; moisture in the solvent or uncontrolled RH is a risk; delayed secondary drying leaves a mobile, solvent-rich product; in thermal processing the cooling rate must be sufficient; a drug loading too high for the polymer leads to crystallization, described by ternary drying phase diagrams (Sadowski and co-workers). - **Kind**: cited (Haser 2017; Dohrn, Sadowski and co-workers; refs 22, 93–100) - **Wiki**: [ASD process F3](../../asd-process/asd-process/findings.md#f3): "Spray drying repeats the spin-coating pattern: dry solvents give homogeneous particles, and water in the spray solvent gives drug-rich domains"; [ASD process F20](../../asd-process/asd-process/findings.md#f20): "A droplet-drying model with a four-component Flory–Huggins phase diagram puts the onset of water-induced separation of ritonavir–PVPVA near 12.5:87.5 water:methanol". - **Verdict**: wiki extends - **Note**: The wiki adds detailed evidence on water during film formation and spray drying (amorphous phase separation rather than crystallization) and on mixed solvents ([ASD process F5](../../asd-process/asd-process/findings.md#f5)); delayed secondary drying and the cooling rate of extrudates are not covered. ## 4. Crystallinity detection methods and in vitro performance testing ### R21. Detection needs orthogonal methods, and XRPD, the most common one, has poor sensitivity - **Review** [p. 7]: detection of crystallinity in ASDs is a significant analytical challenge; orthogonal tools with known limitations should be combined; dilution of the ASD in a drug-product excipient matrix hinders detection, and many tools are unsuited to QC; XRPD has relatively poor sensitivity, limited by mass fraction, dilution, crystal quality and method parameters. - **Kind**: framework, cited (refs 9, 30, 101–104) - **Wiki**: [Characterization methods F4](../../characterization-methods/findings.md#f4): "Each method has a detection limitation or a non-specific signal ... revealed only when the method was compared with a second"; [Solid-state methods F7](../../characterization-methods/solid-state-methods/findings.md#f7): "to 1% or more (XRPD, Raman, DSC)". - **Verdict**: agrees - **Note**: The wiki adds that crystallinity and detection limits are quoted on different bases (drug, powder, volume), so numbers change several-fold with drug loading ([Solid-state methods F4](../../characterization-methods/solid-state-methods/findings.md#f4)); dilution in a tablet is not covered. ### R22. 19F solid-state NMR quantified 0.04% crystallinity in an ASD tablet - **Review** [p. 7]: ssNMR distinguishes small crystalline fractions but is not a common QC tool; 19F ssNMR quantified 0.04% crystallinity in a tablet of a fluorinated drug, a result that may not generalize to other drugs or instruments. - **Kind**: cited (refs 105, 106) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki uses ssNMR only for domain size and homogeneity ([Solid-state methods F20](../../characterization-methods/solid-state-methods/findings.md#f20)), not for quantifying crystals. ### R23. DSC and microscopy: one Tg does not prove homogeneity, DSC misses crystals, PLM, TEM and SEM find crystals that XRPD misses - **Review** [pp. 7–8]: a single Tg can hide inhomogeneity on a smaller length scale; DSC is relatively ineffective for crystallinity in miscible systems because heating dissolves crystals; PLM detects crystals even in X-ray amorphous samples but is diffraction limited; HME crystals are regularly smaller than 1 µm and as small as 5−10 nm; TEM is highly sensitive but low throughput; SEM detects only surface crystals. - **Kind**: cited (refs 8, 9, 12, 30, 43, 62, 75, 102, 107, 114–116) - **Wiki**: [ASD process F10](../../asd-process/asd-process/findings.md#f10): "PLM finds crystals that XRPD misses, DSC finds the fewest, TEM finds nanometre crystals in X-ray-amorphous extrudates"; [Solid state stability F8](../../solid-state-stability/findings.md#f8): "a single DSC glass transition does not exclude it". - **Verdict**: agrees - **Note**: Same ordering, from the same extrusion papers; the wiki adds SEM finding crystals months before XRPD in stored powders (Duong 2022, Moseson 2022). ### R24. SHG quantifies trace crystallinity in the parts-per-billion range; micro-CT detects below 1% but only crystals of about 10 µm or larger - **Review** [p. 8]: SHG, highly sensitive to non-centrosymmetric crystals, enables quantification of trace crystallinity "in the parts per billion range"; X-ray micro-CT identifies crystallinity below 1% but resolves only crystals of about 10 µm or greater. - **Kind**: cited (Schmitt, paclitaxel protein-bound particles; Wanapun 2011; Griffin; Neilly, X-ray microscopy of fenofibrate ASDs) - **Wiki**: [Solid-state methods F7](../../characterization-methods/solid-state-methods/findings.md#f7): "Quoted detection limits for crystals in amorphous powders run from about 10 ppm (SHG, calculated) to 1% or more"; [Solid-state methods F11](../../characterization-methods/solid-state-methods/findings.md#f11): "SHG sees only non-centrosymmetric crystals larger than about 100 nm". - **Verdict**: differs - **Note**: The lowest SHG limit in the wiki is 10 ppm, and that is calculated; the lowest imaged level is 0.1% (naproxen–HPMCAS) and 0.005% (flutamide, ezetimibe), with calculated limits of 0.03–0.047% (Kestur 2012, Correa-Soto 2017). Parts per billion is three orders of magnitude lower; the cited SHG papers (Schmitt on a paclitaxel formulation, Wanapun 2011, Griffin) are not in the wiki, so whether they report such a value can only be settled from them. Micro-CT is not in the wiki. ### R25. Release testing is more sensitive to crystals than most solid-state methods, but QC variability limits it - **Review** [p. 8]: in vitro release has been more sensitive to crystals or phase changes than most other analytical techniques, but a typical variability of ±5% suggests that less than 10% crystallinity may not be detectable in a QC environment; release testing may serve as a challenge test to validate solid-state methods. - **Kind**: cited (Purohit 2018; refs 121–125) - **Wiki**: [Bioperformance F16](../../bioperformance/bioperformance/findings.md#f16): "Compendial tests hardly distinguish crystallinity levels in an amorphous tacrolimus capsule, whereas tests in 100 and 40 mL of medium separate them". - **Verdict**: wiki extends - **Note**: The wiki gives the data behind the statement: in USP test I release fell only from about 93% (10% crystalline) to 58% (100%), and test II showed no trend, while low-volume tests separated the levels (Purohit 2018). ### R26. Non-sink conditions, and an absorptive compartment, are needed to assess supersaturating products - **Review** [p. 8]: traditional sink conditions have little relevance to biopharmaceutical performance of ASDs; non-sink conditions capture release and crystallization kinetics and are essential; an absorptive compartment links formulation changes and release with a more in vivo relevant platform. - **Kind**: framework, cited (refs 80, 130–136) - **Wiki**: [Overview F14](../../overview/findings.md#f14): "Closed-vessel tests misjudged what reaches an absorbing membrane, and the few animal studies agreed with flux or absorptive read-outs more often than with closed-vessel concentration". - **Verdict**: wiki extends - **Note**: The wiki adds the hollow-fibre and diffusion-cell evidence (Hate 2019, 2022; Indulkar 2016) and notes that the library holds exposure data for three drugs in animals and none in humans. ### R27. The sink index against crystalline and amorphous solubility - **Review** [p. 8]: $\mathrm{SI_{cr}} = \frac{C_{cr}}{\mathrm{dose}/V}$ (eq 8) and $\mathrm{SI_{am}} = \frac{C_{am}}{\mathrm{dose}/V}$ (eq 9), with dose the drug added and V the medium volume. - **Kind**: framework, cited (Sun, Wen and Taylor 2016; Sun, Ju and Lee; Schver and Lee 2021) - **Wiki**: [sink index](../../formulas/sink-index.md): "$\mathrm{SI} = \frac{C_s\,V}{\mathrm{Dose}}$", with $C_s$ "solubility (crystalline or amorphous)". - **Verdict**: agrees - **Note**: Algebraically identical. The wiki adds that a crystalline fraction does not dissolve unless the crystalline sink index reaches 1 ([sink index](../../concepts/sink-index.md)). ### R28. The amorphous solubility is the LLPS concentration and depends on the dispersion composition and the medium - **Review** [p. 9]: amorphous solubility is defined as the concentration at which LLPS may be observed, and its value may change with the composition of the drug–polymer system and with media additives. - **Kind**: framework, cited (refs 126, 141–146) - **Wiki**: [amorphous solubility](../../concepts/amorphous-solubility.md): "operationally the concentration at which a drug-rich phase first separates (the LLPS or GLPS onset)"; [Amorphous solubility and LLPS F19](../../dissolution/c-amorphous-solubility-and-llps/findings.md#f19): "Dissolved ASD polymers that enter the drug-rich phase lower the amorphous solubility of ibuprofen". - **Verdict**: wiki extends - **Note**: The wiki adds the thermodynamic definition (metastable equilibrium with the water-saturated supercooled liquid) and quantified shifts: polymers lowering ibuprofen's value from 177.1 to 90.7–96.9 µg/mL (Ueda 2020), HPMCAS-HF lowering four drugs by a quarter to over half ([Bioperformance F19](../../bioperformance/bioperformance/findings.md#f19)), micelles raising it. ### R29. At SIam = 1, crystals in an ASD either only remove potency or grow and cause desupersaturation - **Review** [p. 9]: without crystals the profile is "spring and plateau", or "spring and parachute" if crystallization occurs; with crystals (Figure 7b) either the supersaturation reached is proportional to the amorphous content, the crystals being inert and not growing (low risk), or crystal growth on dissolution lowers the rate and causes desupersaturation, which would translate into a significant loss of bioavailability. - **Kind**: framework, cited (Moseson 2020; Sun, Wen and Taylor 2016) - **Wiki**: [ASD process F29](../../asd-process/asd-process/findings.md#f29): "indomethacin–PVPVA lost mainly potency, bicalutamide–PVPVA lost most of its supersaturation"; the IDM loss is placed "nearest to scenario B (poisoned seeds, only potency lost)". - **Verdict**: agrees - **Note**: The same two scenarios with the same example systems; the wiki adds posaconazole–HPMCAS as a case where 0.3% crystals ended supersaturation sooner (Shah 2024). ### R30. Above the amorphous solubility, drug-rich colloids act as a reservoir that crystals can remove - **Review** [p. 9]: at SIam < 1, the more biorelevant case, concentrations above the amorphous solubility form colloidal drug-rich aggregates by LLPS, which raise bioavailability by a reservoir effect, more colloids keeping the concentration at the membrane near the amorphous solubility; crystals may reduce the achievable supersaturation by growth. - **Kind**: cited (refs 147–152) - **Wiki**: [Bioperformance F5](../../bioperformance/bioperformance/findings.md#f5): "Drug-rich nanodroplets hold the flux at its maximum for a time that grows with the amount of drug in them"; [Supersaturation from ASD dissolution F26](../../dissolution/b-supersaturation-from-asd-dissolution/findings.md#f26): "With crystals already present, a posaconazole dispersion held its amorphous solubility only while undissolved amorphous drug and nanodroplets remained". - **Verdict**: wiki extends - **Note**: The wiki supplies the flux and receiver data (Indulkar 2016; Hate 2018, 2022) and an example of crystals and reservoir acting together (Elkhabaz 2019). ### R31. Detecting crystallinity from the dissolution rate of each solid form has several limitations - **Review** [p. 9]: it needs tight control of particle surface area, works only for the ASD intermediate, not the drug product, applies only to ASDs whose release is governed by the amorphous drug rather than by congruent drug–polymer release, and lacks discrimination because most amorphous compounds dissolve only 2−20× faster than their crystals. - **Kind**: cited (refs 125, 140, 141, 153, 154) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki covers congruent, polymer-controlled release ([Release mechanism F2](../../dissolution/a-release-mechanism-and-loc/findings.md#f2)) but not its use, or failure, as a crystallinity assay; the 2−20× rate ratio is not in the wiki. ### R32. Crystals for performance testing come from three origins, and seed potency follows surface area, not mass - **Review** [p. 9]: grown crystals, spiked (seeded) crystals and residual crystals differ in size, surface area, habit, polymorph and matrix environment; seeds drive growth in relation to their surface area, not their mass, so bulk crystalline material does not mimic intrinsic crystals (Figure 8). - **Kind**: framework, cited (Chasse 2022 industry white paper) - **Wiki**: [crystal seeding](../../concepts/crystal-seeding.md): "the relevant property is the growth area and interface structure of the seeds, not their mass". - **Verdict**: agrees - **Note**: Same principle; the wiki's evidence is paclitaxel seeds of equal mass leaving 7.9–25.3 µg/mL according to their growth area (Que 2018). ### R33. Residual crystals are likely nanometre-sized, of high surface area and defect density - **Review** [p. 10]: after an HME failure residual crystals are most likely the stable form, as nanometre crystals or few large ones; after a spray-drying failure they could be several polymorphs; regardless of origin they are likely below 1 µm with high surface area and defect density. - **Kind**: interpretation, cited (Moseson 2020, 2018a; Moseson 2022) - **Wiki**: [ASD process F26](../../asd-process/asd-process/findings.md#f26): "The authors attribute the extra potency to the small size, embedded position and possible defects of the residual crystals, none of which was measured". - **Verdict**: wiki narrower - **Note**: The wiki records that size, surface area and defect density of residual crystals were inferred rather than measured, apart from TEM domains below 100 nm; the review states them as likely properties. ### R34. Grown crystals sit in a polymer-enriched zone, and undetectable nucleation can still matter - **Review** [p. 10]: grown and residual crystals are in intimate contact with the polymer matrix, which may slow crystallization; grown crystals may have higher polymer concentration nearby where drug was depleted; nucleation events need not be detectable by common methods to have significant performance impact. - **Kind**: interpretation, cited (refs 155–157) - **Wiki**: [Overview F7](../../overview/findings.md#f7): "crystals below the X-ray limit changed performance: SHG, TEM, PLM and SEM found crystals that XRPD and DSC missed". - **Verdict**: agrees - **Note**: The wiki agrees on undetected crystals mattering (bicalutamide with crystals only TEM could find, Moseson 2021); it has nothing on polymer enrichment around grown crystals. ### R35. Spiked crystals underpredict the impact of grown or residual crystals, so weight-percent limits set with them are unreliable - **Review** [p. 10]: spiked crystals from bulk drug are of defined form and size, often larger than 10 µm, with low surface area and defect density and no intimate matrix contact, so they grow more slowly; the same 10% crystallinity from spiked crystals represents far lower surface area, so limits based on spiked crystals likely underpredict, and dissolution is a poor direct surrogate for solid-state quantification; spiked crystals do allow different polymorphs to be tested. - **Kind**: interpretation, cited (refs 6, 10, 124, 158, 159) - **Wiki**: [ASD process F26](../../asd-process/asd-process/findings.md#f26): "Residual crystals in a dispersion lower the supersaturation reached and seed growth more strongly than bulk or added crystals of larger mass". - **Verdict**: wiki extends - **Note**: The wiki shows the direction in two systems (bicalutamide relative form 1 growth 0.088 and 0.071 for residual crystals against 0.044 for 10% external seeds; Moseson 2021) and in paclitaxel (Que 2018), but notes a different comparison for tacrolimus in dogs (see the last section). ## 5. Case studies ### R36. Indomethacin–PVPVA extrusion: below Tc residual crystals always remain, and detection depends on the method - **Review** [pp. 10–11]: Tc of 131 °C at 50% drug loading; at or above Tc a fully amorphous extrudate formed given sufficient residence time; below Tc XRPD found residual crystallinity whatever the temperature or residence time; PLM was more sensitive than XRPD and both more than DSC in some formulations; in a sample made at Tc for the longest residence time and amorphous by PLM, XRPD and DSC, TEM found crystals of 5−10 nm. - **Kind**: cited (Moseson and Taylor 2018; Moseson 2018a) - **Wiki**: [ASD process F9](../../asd-process/asd-process/findings.md#f9): "Below the critical temperature no extrudate was crystal-free"; "A separate 1:1 extrudate made at 131 °C for 20 min was X-ray amorphous with a single Tg yet held nanometre crystals by TEM". - **Verdict**: agrees - **Note**: The wiki notes that the measured Tc is 132.3 ± 0.8 °C and 131 °C is the Flory–Huggins prediction ([ASD process F8](../../asd-process/asd-process/findings.md#f8)), and gives the crystallinity at each temperature and residence time. ### R37. Indomethacin–PVPVA dissolution: residual and spiked crystals lower supersaturation in proportion to crystal content, and the seeds do not grow - **Review** [p. 11]: non-sink profiles of extrudates with 0−25% residual crystallinity and of physical mixtures with 0−40% spiked crystals fell along the theoretical maximum set by the amorphous content, which "indicates" that the seeds do not grow and dissolution rates are unaffected, polymer adsorption preventing or slowing growth; the extrudates were slightly lower, hypothesized to come from the greater surface area and energy of residual crystals or from XRPD underestimating their crystallinity. - **Kind**: cited (Moseson 2020) - **Wiki**: [ASD process F26](../../asd-process/asd-process/findings.md#f26): "Residual crystals in a dispersion lower the supersaturation reached and seed growth more strongly than bulk or added crystals of larger mass"; extrudates "peaked lower and fell". - **Verdict**: differs - **Note**: Both rest on Moseson 2020. The wiki records the 7–25% extrudates 4–6 µg/mL below the line, slight desupersaturation above 3%, and, at equal amorphous dose, extrudates falling (e.g. 117-2 from about 32.5 to 27.0 µg/mL) where bulk seeds did not, read by the authors as residual crystals "better able to grow"; the review's own section 6.1.5 (p. 17) states the same. The case-study wording "do not grow" holds only for the spiked mixtures, so the wiki's reading is better supported. ### R38. Bicalutamide–PVPVA: crystal-free extrudates crystallize in their matrix to the metastable form - **Review** [p. 11]: Tc of 145 °C; a fully amorphous extrudate and three with crystallinity up to 30%, one below 1% (seen by TEM, not XRD); the crystal-free extrudate showed incomplete release without desupersaturation, with nucleation at the solid−liquid interface by PLM, forming the metastable polymorph. - **Kind**: cited (Moseson 2021) - **Wiki**: [ASD process F27](../../asd-process/asd-process/findings.md#f27): "crystal-free bicalutamide–PVPVA never exceeded about 38–40 µg/mL against an amorphous solubility of 150 µg/mL"; "Raman showed the amorphous solid turning to the metastable form 2". - **Verdict**: agrees - **Note**: Same system and observations; the wiki adds that HPMCAS, a strong solution inhibitor, delayed this route by only about 10 min. ### R39. Bicalutamide–PVPVA: below 1% stable-form crystals are highly detrimental, even near sink, and loading and mechanical activation raise matrix crystallization - **Review** [p. 11]: samples with crystallinity reached distinctly lower supersaturation by rapid growth of the stable polymorph within the amorphous solid; matrix crystallization occurred with crystals present even under sink conditions, showing the detrimental impact of even <1% crystallinity; risk factors for matrix crystallization were drug loading above the limit of congruency and processing that mechanically activated the sample. - **Kind**: cited (Moseson 2021, 2022) - **Wiki**: [ASD process F28](../../asd-process/asd-process/findings.md#f28): "Trace residual form 1 crystals, below the X-ray detection limit, remove the lag, switch the crystallising form from form 2 to form 1 and lower the supersaturation reached at every dose"; [ASD process F38](../../asd-process/asd-process/findings.md#f38), [F39](../../asd-process/asd-process/findings.md#f39). - **Verdict**: wiki extends - **Note**: The wiki gives the numbers (4 h concentration at 150 µg/mL about 35 µg/mL crystal-free against 24.5 with <1% crystals; near sink 1.35–1.92 against 2.45 µg/mL), the 5–10% loadings that released congruently, and cryomilling dropping release to about the form 2 solubility; it notes that the limit of congruency of this pair lies between 10 and 20% but was not located. ### R40. Naproxen–PVP: spray drying left more crystals than HME at 60% loading, and both grew in storage - **Review** [p. 11]: at 30% loading both methods gave amorphous ASDs with no crystallization after 8 weeks at 40 °C; at 60% loading crystallinity after manufacture was 6% (spray dried) and 0.4% (HME), rising to 22% and 3% on storage, read as a loading beyond what the polymer can stabilize and higher drug mobility or driving force during solvent drying than during melt cooling. - **Kind**: cited (Haser 2017) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki has no direct comparison of spray drying and HME for the same composition. ### R41. Kinetisol: higher shear lowers crystallinity but raises degradation, and a threshold balances the two - **Review** [pp. 11–12]: a shear-labile drug (LY3009120) processed by Kinetisol gave degradation of 0.7, 2.0 and 3.1% with crystallinity of 2.3, 0.9 and 0.1%; no change in 4 months at 25 °C/60% RH; 19F ssNMR was needed because DSC and XRPD lacked sensitivity; non-sink pH-shift dissolution showed desupersaturation only at 2.3% crystallinity. - **Kind**: cited (Davis 2021) - **Wiki**: [ASD process F35](../../asd-process/asd-process/findings.md#f35): "The extrusion temperature window has a lower limit set by residual crystals and an upper limit set by chemical degradation". - **Verdict**: wiki narrower - **Note**: The same trade-off is in the wiki for HME of bicalutamide–PVPVA (about 5% degradation at the temperature that removed crystals); Kinetisol, this drug and a performance-based crystallinity threshold are not covered. ### R42. Fenofibrate tablets: spiked and recrystallized API give different dissolution, depending on the method - **Review** [pp. 12–13]: transmission Raman with chemometrics produced tablets with set levels of crystallinity recrystallized during accelerated storage; with two sink methods (SDS- or CTAB-based media, different stirring and temperatures) profiles for 0−40% crystallinity were nearly indistinguishable or very different according to method and crystal source; spiked crystals poorly represent crystals formed in the formulation. - **Kind**: cited (Theil 2018) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The principle is covered by R35, but this study, transmission Raman quantification and the medium dependence of sensitivity in sink tests are not in the wiki. ### R43. Tacrolimus capsules: low-volume tests reveal crystallinity, and fully crystallized product still exceeds the crystalline solubility and outperforms a crystalline suspension in dogs - **Review** [pp. 13–14]: Prograf did not crystallize, one generic crystallized readily and was fully crystallized at elevated temperature/humidity; at the highest sink index (largest volume per dose, 450 mL) up to 20% crystallinity showed no marked loss of supersaturation, at the lowest (40 mL) every crystal-containing sample lost supersaturation and desupersaturated; the 100% crystallized product exceeded the crystalline solubility at both volumes and in beagle dogs outperformed a crystalline suspension in AUC and Cmax, possibly because crystallization in the matrix was incomplete, the crystals were highly defective, or they were very small and redissolved quickly. - **Kind**: cited (Trasi 2017; Purohit 2018, 2019) - **Wiki**: [Bioperformance F16](../../bioperformance/bioperformance/findings.md#f16): "tests in 100 and 40 mL of medium separate them"; [Bioperformance F30](../../bioperformance/bioperformance/findings.md#f30): "lost about half of its exposure but gave about four times the exposure of a suspension of reference crystals". - **Verdict**: wiki extends - **Note**: The wiki gives the numbers (AUC0–24 42.1, 21.6 and 5.29 ng·h/mL for fresh, 100% crystallized and reference crystals), crystals of 1–5 µm in the capsule against 5–25 µm reference crystals, which supports the review's third explanation, release above 1.8 µg/mL at every volume, and the absorptive and simulated results ([F28](../../bioperformance/bioperformance/findings.md#f28), [F17](../../bioperformance/bioperformance/findings.md#f17)). The wiki's 450 mL summary states only that release at 120 min fell with crystallinity, without comparing 0 and 20%. ### R44. Miscible dispersions resist crystallization and immiscible ones crystallize within months; the matrix controls whether crystals grow - **Review** [pp. 13–14]: solvent-evaporated dispersions classed as miscible were stable at ambient conditions and immiscible ones crystallized within 1−2 months; nifedipine–PVP at 70% loading, classed miscible, had initial residual crystals that grew for 6 months and then stopped below about 10%, attributed to drug not stabilized by the polymer; ketoconazole–PVP was miscible at 30% (no crystals in about 1 year) and immiscible at 70% (crystals after about 2 months, progressing). - **Kind**: cited (Ivanisevic 2010) - **Wiki**: [Physical stability F30](../../solid-state-stability/physical-stability/findings.md#f30): "Dry indomethacin dispersions crystallized within days above the drug loading at which a drug-rich phase appeared and stayed amorphous for more than a year below it". - **Verdict**: wiki extends - **Note**: The wiki has the same relation for indomethacin with HPMC and HPMCAS (Tian 2016) and for water-separated systems ([Physical stability F15](../../solid-state-stability/physical-stability/findings.md#f15)); Ivanisevic's study is not in the wiki, and the wiki leaves the phase state of ketoconazole–PVP unresolved (see the last section). ## 6–7. Assessment of critical quality attributes and conclusions ### R45. Risk of crystallinity is assessed in five categories, as scientific rationale rather than a workflow - **Review** [pp. 14–15]: drug physicochemical properties (polymorphs, crystallization tendency, Tm, Tg, amorphous/crystalline solubility ratio), ASD formulation fundamentals, physical stability, dissolution considerations and crystal origin and micromeritic properties (Figure 16); the pyramid is not a sequence or decision tree. - **Kind**: framework - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki's overview findings cover the same factors one by one (e.g. [Overview F4](../../overview/findings.md#f4) on the neat-drug class as a first flag) but have no crystallinity risk framework. ### R46. A metastable polymorph can keep some supersaturation, but stable-form crystals in the product remove that route - **Review** [p. 15]: drugs without known crystalline forms (ledipasvir, clopidogrel) carry no crystallinity risk; if a supersaturated solution crystallizes first to a metastable polymorph, its higher solubility maintains some supersaturation over the stable form; in bicalutamide, stable-polymorph crystals in the formulation removed the natural route to the metastable form, which was highly detrimental. - **Kind**: interpretation, cited (Moseson 2021) - **Wiki**: [ASD process F28](../../asd-process/asd-process/findings.md#f28): residual form 1 crystals "switch the crystallising form from form 2 to form 1 and lower the supersaturation reached at every dose". - **Verdict**: agrees - **Note**: Same mechanism; the wiki also records metastable forms from melts and in dissolution (nifedipine γ before α, [Supersaturation from ASD dissolution F3](../../dissolution/b-supersaturation-from-asd-dissolution/findings.md#f3)). The absence of crystal forms for ledipasvir and clopidogrel is not stated in the wiki. ### R47. Crystallization tendency classes and mobility-based predictors flag risk but do not guarantee stability - **Review** [p. 15]: classification systems, often from a melt cooling/reheating cycle, categorize drugs; molecular mobility and configurational entropy (ΔCp at Tg) have been used to predict amorphous stability; a high-Tg polymer and drug–polymer interactions do not guarantee stability, and Newman and Zografi add the preparation method, heterogeneous nucleation from processing, Johari−Goldstein relaxations, non-diffusional growth in the glass and surface crystallization. - **Kind**: cited (Baird 2010; Newman and Zografi; refs 69, 163–167) - **Wiki**: [Crystallization tendency F17](../../solid-state-stability/crystallization-tendency/findings.md#f17): "The neat-drug class is a first risk flag for keeping a drug amorphous, in storage and in polymer dispersions, not a prediction"; [Solid state stability F2](../../solid-state-stability/findings.md#f2): "No bulk glass or thermodynamic property ranked crystallization, for neat drugs or for dispersions". - **Verdict**: wiki extends - **Note**: The wiki adds that reduced Tg, fragility and glass-stability parameters fail to rank, and gives data for surface crystallization and processing-derived nuclei ([Solid state stability F11](../../solid-state-stability/findings.md#f11)); Johari−Goldstein relaxations are not covered. ### R48. A miscible, high-Tg, low-hygroscopicity polymer at low loading with interactions lowers the risk, also in the presence of seeds - **Review** [p. 15]: reduced risk of physical stability failure comes with a miscible, high-Tg polymer of low hygroscopicity, lower drug loading (a thermodynamic and kinetic effect) and drug–polymer interactions that slow crystallization; then even residual crystal seeds should grow slowly. - **Kind**: framework, cited (refs 19, 67, 74, 90) - **Wiki**: [Solid state stability F6](../../solid-state-stability/findings.md#f6): "A polymer protects best when it has a group complementary to a free group of the drug, and the best polymer reverses with the drug"; [Physical stability F11](../../solid-state-stability/physical-stability/findings.md#f11): "In humid storage the hygroscopic polymer loses its advantage". - **Verdict**: wiki extends - **Note**: The wiki agrees on each factor and adds that the best polymer depends on the drug and the stress; it has no storage data on seeded or residual-crystal-containing dispersions to test the last clause. ### R49. In dissolution, strong drug–polymer interactions can impede release; seeds can trigger solution or matrix crystallization; lower loading reduces the matrix route - **Review** [p. 15]: the polymer can facilitate dissolution and prolong supersaturation, but strong drug–polymer interactions that persist in water may impede release; seeds can facilitate solution-mediated or solid-mediated (matrix) crystallization with widely varying impact; polymer adsorption can poison growth; supersaturation modulates the driving force; matrix crystallization is reduced by lowering drug loading. - **Kind**: interpretation, cited (Moseson 2020, 2021, 2022; refs 74, 168) - **Wiki**: [Release mechanism F17](../../dissolution/a-release-mechanism-and-loc/findings.md#f17): "In PVPVA, drugs that hydrogen-bond to the polymer had lower limits of congruency than close analogues that cannot"; [ASD process F38](../../asd-process/asd-process/findings.md#f38): "5–10% loading released drug and polymer together without matrix crystallization". - **Verdict**: wiki extends - **Note**: The wiki supports each clause and adds the limit of congruency as the loading threshold for congruent release (indomethacin 10% against at least 25% for its methyl ester; Saboo and others). ### R50. Surfactants lower Tg, change phase-separation morphology and can promote crystallization in solution - **Review** [p. 15]: surfactant inclusion is expected to reduce physical stability by lowering Tg, can induce various phase-separation morphologies that influence crystallization, and can facilitate dissolution but also induce solution-phase crystallization. - **Kind**: cited (refs 116, 169–180) - **Wiki**: [Physical stability F22](../../solid-state-stability/physical-stability/findings.md#f22): "Surfactants that mix with the amorphous drug speed its crystal growth"; [Dissolution F6](../../dissolution/findings.md#f6): "The surfactants that raise the limit of congruency also promote crystallization". - **Verdict**: wiki extends - **Note**: The wiki quantifies it (TPGS raised celecoxib growth about 85-fold at 70 °C and lowered Tg about 10 °C, SLS did not; Mosquera-Giraldo 2014) and records the round versus other domain shapes surfactants give humid ritonavir films ([Physical stability F32](../../solid-state-stability/physical-stability/findings.md#f32)). ### R51. Water is a third component that shifts both the solubility line and Tg, so humidity can turn stable compositions metastable and unstabilized - **Review** [p. 16]: PC-SAFT diagrams for naproxen with PVP, PVPVA and HPMCAS at 0, 60 and 75% RH show the solubility line moving to lower drug contents with humidity (40% loading becoming more unstable, 20% becoming metastable at 75% RH for PVP and PVPVA); dry Tg − T of 10 to 100 °C falls to −5 to 25 °C at 60% RH, and at 75% RH several compositions lose kinetic stabilization (against the rule Tg − T > 50 °C); packaging against humidity matters. - **Kind**: cited (Lehmkemper, Sadowski and co-workers 2017) - **Wiki**: [Physical stability F6](../../solid-state-stability/physical-stability/findings.md#f6): "Dispersions take up more water than the amorphous drug ... and the water erases the Tg gain that the polymer gives when dry". - **Verdict**: wiki narrower - **Note**: The wiki has the Tg side with data (all systems near 28–36 °C at 75–76% RH; Konno 2008, Marsac 2008) and water-induced separation, but no calculated shift of the crystalline solubility line with humidity. ### R52. Subambient storage may suit systems with high crystallization risk - **Review** [p. 16]: storage temperature changes both kinetic stability (Tg − T) and relative supersaturation, so for some high-risk systems subambient temperatures may be the most appropriate long-term storage condition. - **Kind**: interpretation - **Wiki**: [Crystallization tendency F4](../../solid-state-stability/crystallization-tendency/findings.md#f4): "Nucleation can be fastest far below Tg, so cold storage does not protect every drug glass". - **Verdict**: differs - **Note**: The review's suggestion rests on lower mobility and lower supersaturation; the wiki's nilutamide glass nucleated most at −30 °C (about 60 °C below Tg), favouring a metastable form (Trasi 2014). The difference comes from different systems (an ASD in general against one neat drug glass), but it shows that subambient storage needs a nucleation check before it is chosen. ### R53. Crystals grow only where the matrix is supersaturated, at a rate set by Tg − T - **Review** [p. 16]: in a matrix that is not supersaturated the crystals will not grow; in metastable systems Tg − T sets the growth rate; in the nifedipine–PVP case growth stopped when the amorphous drug's solubility in the polymer was presumably reached. - **Kind**: interpretation, cited (Ivanisevic 2010) - **Wiki**: [Physical stability F15](../../solid-state-stability/physical-stability/findings.md#f15): felodipine–PVP powders "stopped within about 20 days at plateaus of about 70, 80 and 75%, part of the drug staying amorphous for 494–724 days". - **Verdict**: wiki narrower - **Note**: The wiki records crystallization plateaus in water-separated felodipine–PVP but does not attribute them to the drug's solubility in the polymer, and has no residual-crystal growth data in storage to test the Tg − T statement. ### R54. Test design and media decide whether crystallinity is detected: two-stage pH tests and bile salt/lecithin composition - **Review** [p. 17]: a QC sink test is unlikely to detect crystallinity; for enteric-polymer ASDs gastric pH changed matrix crystallization (pH 1.6 good release, pH 3 incomplete release, single stage at pH 6.5 complete), so a pH 3 to 6.5 shift would discriminate better; nucleation induction times can rise or fall in complex media; lecithin and bile salt levels alter crystallization, and in vivo bile salts are more diverse than sodium taurocholate in commercial media. - **Kind**: cited (Nguyen 2023; refs 88, 183, 187–191) - **Wiki**: [Supersaturation from ASD dissolution F25](../../dissolution/b-supersaturation-from-asd-dissolution/findings.md#f25): "gastric pH 3.0 or 5.0 was often worse than pH 1.6 also for neutral praziquantel, and a single stage at pH 6.5 ranked dispersions alike that later failed"; [Crystallization from solution F40](../../dissolution/d-crystallization-from-solution/findings.md#f40): "a medium with the six main human bile salts held atazanavir, ritonavir, tacrolimus and telaprevir supersaturated about two- to more than fortyfold longer". - **Verdict**: wiki extends - **Note**: The wiki has the delamanid data the review summarizes, adds praziquantel (Alvarenga 2025), and quantifies lecithin and its digestion products shortening induction times ([D F39](../../dissolution/d-crystallization-from-solution/findings.md#f39)). ### R55. An HPMCAS-HF dispersion showed a negative food effect because the polymer does not dissolve below about pH 6.8 - **Review** [p. 17]: a negative food effect was observed for an HPMCAS-HF ASD; the polymer does not ionize and dissolve until ≥ pH 6.8, release differed across biorelevant media, and fed-state GI pH below the polymer threshold reduced bioavailability. - **Kind**: cited (Nguyen 2023, pretomanid) - **Wiki**: [Bioperformance F26](../../bioperformance/bioperformance/findings.md#f26): "fell to 72% of its fasted exposure with food while the crystalline product rose to 163%; in vitro the dispersion released only to the crystalline solubility below FeSSIF pH 5.8". - **Verdict**: wiki extends - **Note**: Same study; the wiki adds the monkey numbers and that released particles showed crystalline pretomanid, linking the food effect to crystallization. It also records release of about 143 µg/mL at FeSSIF pH 6.0, which places the practical release threshold in fed media below the nominal pH 6.8. ### R56. Representative crystal-containing samples can be made by triggering the failure mode - **Review** [p. 17]: for spray drying, an antisolvent such as water in the spray solvent can trigger phase separation and/or crystallization; for HME, lower temperatures, shorter residence times or larger input crystals induce residual crystallinity; Hermans used milled and unmilled API, which changed the sensitivity of dissolution to crystals, and residual crystals gave more desupersaturation than added seeds. - **Kind**: framework, cited (Moseson 2018, 2020, 2022; Hermans; refs 93, 95) - **Wiki**: [ASD process F9](../../asd-process/asd-process/findings.md#f9): "at and above it residence time and input crystal size decided"; [ASD process F3](../../asd-process/asd-process/findings.md#f3): "water in the spray solvent gives drug-rich domains or a drug-enriched particle surface". - **Verdict**: agrees - **Note**: The wiki's transfer checks propose the same deliberate extrusion of lots at the extremes of a size specification; Hermans' milled and unmilled API study is not in the wiki. ### R57. A high-Tg amorphous matrix may limit growth of residual crystals - **Review** [p. 17]: even with residual crystals, matrix properties such as a high glass transition temperature may reduce the risk of crystal growth. - **Kind**: interpretation - **Wiki**: [Physical stability F5](../../solid-state-stability/physical-stability/findings.md#f5): "Dispersion Tg, Tg-scaled temperature and bulk mobility measures do not rank polymers or drugs by crystallization rate". - **Verdict**: wiki narrower - **Note**: The wiki's evidence (unseeded nucleation and growth) shows that dispersion Tg alone did not rank crystallization, and a low-Tg polymer could speed growth ([Physical stability F28](../../solid-state-stability/physical-stability/findings.md#f28)); the review's statement is more general than either source tests for residual seeds. ### R58. Amorphous–amorphous phase separation must also be assessed, and is harder to detect than crystals - **Review** [p. 17]: demixing into drug-rich and polymer-rich domains, from poor miscibility at high loading or water uptake, is far harder to detect by PXRD or PLM and can significantly affect dissolution and bioavailability. - **Kind**: framework, cited (ref 192) - **Wiki**: [Overview F9](../../overview/findings.md#f9): "Water separates drug–polymer dispersions at every stage ... but a separation made before dissolution raised release in one system and lowered it in another". - **Verdict**: wiki extends - **Note**: The wiki confirms the detection difficulty (one DSC Tg with domains present, [Solid state stability F8](../../solid-state-stability/findings.md#f8)) and adds a case where deliberate separation raised release (lopinavir–HPMC, [ASD process F14](../../asd-process/asd-process/findings.md#f14)), so the effect is not only negative. ### R59. Downstream processing to a drug product can induce demixing or crystallization and dilutes the ASD for detection - **Review** [p. 17]: milling and compression are high-energy operations that can induce demixing or crystallization; immediate-release tablets need high disintegrant levels while some products (ritonavir) are erodible tablets; release depends on ASD loading, polymer and tablet formulation; dilution in excipients complicates detection. - **Kind**: cited (refs 160, 193–200) - **Wiki**: [ASD process F39](../../asd-process/asd-process/findings.md#f39): "Cryomilling a 30% bicalutamide–PVPVA dispersion for 2–5 min lowered the concentration reached to about the form 2 solubility"; [ASD process F48](../../asd-process/asd-process/findings.md#f48): "Pressing coated ritonavir–PVPVA particles into a tablet lowered release at pH 6.8". - **Verdict**: wiki narrower - **Note**: The wiki covers milling and compression in two systems; disintegrant level, erodible tablets and the effect of excipient dilution on crystallinity detection are not treated. ### R60. Dissolution as a surrogate specification test has limited power; specifications need a product-specific risk assessment - **Review** [pp. 17–18]: ICH Q6A treats surrogate tests such as dissolution as typically suitable and solid-state methods as a last resort, but QC dissolution and content uniformity lack the specificity to quantify crystallinity, sink conditions do not discriminate solubility-driven differences, and a statistical analysis found about 6% crystalline content detectable only under ideal assumptions; a non-sink QC test should be applied only when a risk assessment (development studies, PC-SAFT, IVIVC) justifies it, or sensitive solid-state methods linked to bioperformance used instead. - **Kind**: interpretation, cited (ICH Q6A; Hermans) - **Wiki**: none - **Verdict**: wiki silent - **Note**: The wiki's transfer check in ASD process advises not to accept "below the XRPD limit of detection" as a specification for a fast-crystallizing, low-Tg drug, but it has no finding on regulatory specification or on statistical detection limits of QC dissolution. ### R61. There is no common acceptable crystallinity level: in some systems trace crystals are harmless, in others they are highly detrimental - **Review** [p. 18]: crystallinity is a complex CQA because detection is difficult and acceptable levels differ between drug–polymer systems; in some, growth is slow and low levels hardly affect supersaturation, while in others X-ray amorphous samples crystallize rapidly from trace crystals; each product should be assessed by studies designed to identify its failure modes, aiming at in vitro−in vivo correlations. - **Kind**: interpretation - **Wiki**: [ASD process F29](../../asd-process/asd-process/findings.md#f29): "Whether residual crystals are tolerable depends on the system". - **Verdict**: agrees - **Note**: Same conclusion, with indomethacin–PVPVA (mainly potency lost) against bicalutamide–PVPVA and posaconazole–HPMCAS (supersaturation lost from below 1% or 0.3% crystals). ## Wiki statements the review would correct or qualify Applied on 2026-11-04 with the user's agreement; the quotes below are the wording before that date. See `review/corrections_review.html`. - [Bioperformance F30](../../bioperformance/bioperformance/findings.md#f30) — "so spiking with purchased crystals would have overstated the loss" — The review uses "spiked" for bulk crystals mixed into an amorphous product and argues they underpredict the loss (p. 10); the tacrolimus comparison in F30 is a suspension of reference crystals against a fully crystallized capsule, not a spiked product, which the review describes as needing a representative crystalline reference (p. 14). The title wording mixes two different comparisons. - [Miscibility F4](../../solid-state-stability/miscibility-and-drug-polymer-interactions/findings.md#f4) — "Ketoconazole ... shows one Tg at every ratio, but whether it forms one phase is unresolved" — The review cites Ivanisevic (2010) classifying ketoconazole–PVP as miscible at 30% and immiscible at 70% loading, with crystallization only at 70% (pp. 13–14); this is outside evidence on the open question, though not from the wiki's papers. - [Crystallization tendency F4](../../solid-state-stability/crystallization-tendency/findings.md#f4) — "the '50 °C below Tg' rule sits at the edge of the deep-cold nucleation zone for this drug" — The review notes that several ASDs reached years of physical stability under much less stringent conditions than Tg − 50 °C (p. 6), so the rule is a conservative guide for ASDs as well as a non-guarantee for neat glasses. - [formulation critical temperature](../../concepts/formulation-critical-temperature.md) — "for 1:1 IDM–PVPVA it is 131 °C" — The review also uses 131 °C (p. 10); as the wiki itself records in F8, this is the Flory–Huggins prediction and the measured value is 132.3 ± 0.8 °C, so the concept definition should name 131 °C as predicted. ## Topics of the review that the wiki does not cover - Regulatory basis of crystallinity specifications (ICH Q6A, Q8(R2)) and the statistical detection limit of QC dissolution (about 6%, Hermans) — pp. 1, 8, 17–18 — sets whether dissolution or a solid-state method can serve as the release test for crystalline content. - Kinetisol, microwave and mechanochemical amorphization, and the method classes of ASD manufacture — pp. 3, 11–12 — alternative routes have their own residual-crystal and degradation trade-offs. - Alternative routes to Tc and phase diagrams (recrystallization method, PC-SAFT, empirical models, rheology) — p. 4 — needed where DSC melting offsets are unreliable. - 19F ssNMR (0.04% crystallinity in a tablet), micro-CT and vibrational or terahertz methods for crystallinity — pp. 7–8 — the only quoted route to sub-0.1% quantification in a finished tablet. - Crystallinity detection from form-specific dissolution rates and its limits (2−20× rate ratio) — p. 9 — explains why rate-based assays fail for many ASDs. - Spray drying against HME for the same composition (naproxen–PVP, Haser 2017) — p. 11 — shows the process route changes initial crystallinity at the same loading. - Spiked against recrystallized API in tablets, with medium-dependent sensitivity (fenofibrate, Theil 2018) — pp. 12–13 — direct evidence on reference materials in sink tests. - Humidity-dependent PC-SAFT phase diagrams (naproxen with three polymers) — p. 16 — links packaging and storage RH to both thermodynamic and kinetic stability. - Drug-product design: disintegrant levels, erodible tablets and dilution of the ASD in excipients — p. 17 — governs detection and release of the marketed dosage form.