When weakly basic drugs go back onto the silica: pH, surface charge and release from mesoporous silica formulations

Background and motivation

Mesoporous silica is a carrier for poorly soluble drugs. A drug loaded into pores a few nanometres wide cannot crystallize, especially when the pores are smaller than the critical nucleus, so it is held in an amorphous state (refs 3–5). On contact with water such formulations often release the drug in a burst and create a supersaturated solution, which raises the driving force for absorption across the gut wall (Taylor and Zhang 2016, ref 1; Mellaerts 2008, ref 9; Dening and Taylor 2018, ref 10). The authors abbreviate mesoporous silica-based formulations as MPS. C1·C2·C3

Silica is not an inert container, though. Its surface carries silanol groups (isolated, vicinal and geminal), which bind guest molecules, most often by hydrogen bonds; isolated silanols are the most reactive sites (refs 13, 14). Depending on the drug and the pH, binding can also be electrostatic (Song 2005, ref 16). Strong drug–silica binding gives high loading but sustained, often incomplete release; weak binding gives lower loading and a burst (refs 17, 18). This is why surface functionalization is used to tune release: grafting amine groups moves the isoelectric point of silica to about pH 7, so that weakly acidic drugs are held at higher pH, and a triethylammonium-modified silica with an isoelectric point of pH 7.38 was used to deliver sulfasalazine to the colon (refs 16, 18–23, 26). C4·C6·C7

The charge of plain silica depends on pH. Its isoelectric point lies near pH 2: below it, the silanols carry an extra proton (≡SiOH₂⁺); above it, they lose one (≡SiO⁻) and the surface becomes increasingly negative (Colilla 2010, ref 15; Sjöberg 1996, ref 44). The drug's charge also depends on pH. For weak acids this combination is favourable at intestinal pH: ibuprofen, indomethacin and glibenclamide released better from silica at higher pH, explained by higher solubility and by repulsion between the anionic drug and the anionic silica (refs 24, 25). C5·C8·C25

Weak bases are the opposite case and had received little attention. A weak base dissolves well in the acidic stomach and becomes supersaturated when the pH rises past its pKa in the small intestine. For an MPS formulation, supersaturation can increase the amount of drug adsorbed on the silica and so lower the dissolved drug available for absorption (Parfitt and Rochester 1983, ref 27). Earlier work showed one consequence of the pH shift: itraconazole (pKa 3.7) released from ordered mesoporous silica crystallized after the shift (Mellaerts 2008, ref 9). The authors' own earlier atazanavir work (a 2020 thesis, ref 28) found that starting dissolution in acid improved release from silica, that a polymer in the medium could hold the supersaturation after the shift, and that an absorptive compartment led to almost complete release. C9·C10·R23

What had not been reported is whether a weak base that has dissolved in acid could re-adsorb onto the silica when the pH rises, simply because it becomes less soluble while the silica becomes more negative. The authors call this "a possibility" and set out to probe how pH governs the balance between adsorption and release of poorly soluble weak bases under supersaturated conditions. They chose three weak bases with different pKa values so that different kinds of drug–silica interaction could be compared: ketoconazole (KET, pKa 2.9 and 6.5), clozapine (CLZ, pKa 7.5 in the introduction and 7.3 in the rest of the paper) and atazanavir (ATZ, pKa 4.49). Each was loaded at 20% w/w into SBA-15 and tested by adsorption isotherms, single-pH and two-step (pH-shift) release, competitor experiments and, for KET, release with an absorptive membrane. C11·C13·M1·M2·M3·X1

Background concepts

pKa and the fraction ionized. A weak base B takes up a proton to form the cation BH⁺. The pKa is the pH at which half of the drug is protonated. Below the pKa the drug is mostly charged; one pH unit above it, only about one molecule in ten is still charged. The authors calculated the percent ionized for each drug with the Henderson–Hasselbalch equation (named, not printed, in the paper) and plotted it in Figure 1. KET has two basic sites, so it exists as BH₂²⁺, BH⁺ or B; at pH 4.5 it is almost entirely BH⁺. Read from Figure 1, the fraction still charged at pH 6.8 is about 33% for KET, about 76% for CLZ and about 0.5% for ATZ; at pH 5.5 it is about 91%, 98% and 9%. R1·E1

pH-dependent solubility. The neutral form of a poorly soluble base has a low intrinsic solubility; the charged form is far more soluble. Below the pKa the total solubility therefore rises about tenfold for each pH unit. Figure S4 shows the calculated crystalline pH-solubility profiles: each curve leaves its low plateau near pH 5 (ATZ), 7 (KET) and 8 (CLZ). The unit of the plotted log(solubility) is not given. R19

Zeta potential and isoelectric point. The zeta potential is the electrical potential at the slipping plane of a particle in liquid, measured from how fast the particle moves in an electric field. Its sign gives the sign of the net surface charge; the isoelectric point is the pH at which it crosses zero. P4

Adsorption isotherm. An adsorption isotherm plots the amount of solute held on a solid (here mg of drug per g of SBA-15) against the concentration left in solution at equilibrium, at constant temperature. The authors measured it by solution depletion: the adsorbed amount is the difference between the drug in solution at the start and at equilibrium. They fitted the isotherms with the BET equation (named, not printed) to obtain a monolayer coverage Q_m, the amount needed to cover the surface with one layer of molecules (refs 28, 43). P2·P14·E2·E4

Solute activity. To compare drugs with very different solubilities, Figure S5 replots the pH 6.8 isotherms against solute activity, defined in the SI as the equilibrium concentration after adsorption divided by the amorphous solubility, $a=C_{\mathrm{e}\mathrm{q}}/S_{\mathrm{a}\mathrm{m}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{h}}$. In plain terms: an activity of 1 means the solution is as concentrated as it can be before an amorphous drug phase would form, so the activity measures how far the solution is from that limit. E3·P15

Two-step (pH-shift) dissolution. The formulation is first suspended in acid (here 0.1 N HCl, about pH 1, for 30 min), then the pH is raised to an intestinal value by adding concentrated NaOH. This mimics gastric emptying in the fasted state. A fed-state version starts at pH 4.5 and shifts to 5.5. P6

Absorptive dissolution. A dissolution vessel (donor) is connected to a membrane through which dissolved drug passes into a receiver solution, as drug would pass into the gut wall. Removing dissolved drug can shift competing processes such as crystallization or adsorption (refs 35–37). The apparatus here uses a cellulosic hollow-fiber module. P8·C12

Competitive adsorbates. To find out how a drug binds, a second small molecule that binds the surface in a known way is added. Triethylamine (TEA) is a base that is positively charged at pH 6.8 and competes for negative silanolate sites; urea competes for hydrogen-bond sites (refs 42, 43). If TEA removes the drug from the surface and urea does not, the drug is held mainly by electrostatic attraction. C19·P3

Methods in detail

The study combines five kinds of measurement on the same three drugs and the same silica. The flow below shows how they build on each other: the charges of silica and drug come first, then adsorption and release at fixed pH, then the pH shift, and finally two tests that probe why dissolved drug disappears after the shift.

Formulations and solid state

KET and CLZ (both from Hawkins, Inc.) and ATZ (Gojira Fine Chemicals) were loaded into SBA-15 from Glantero (pore size 7.1 nm, specific surface area 586 m²/g) by incipient wetness impregnation: a concentrated methanolic stock (20–40 mg/mL) was added to the silica to give 20% w/w drug, mixed with a spatula, dried at 40 °C overnight and then under vacuum for 48 h. The drug content was checked by extracting a known amount in methanol (30 min sonication) and analysing the supernatant by HPLC; the measured contents are not reported. PXRD of the three formulations shows only a broad halo near 22–23° 2θ (Figure S1), and DSC shows no glass transition, crystallization or melting (Figure S2), which the authors read as confinement of the drug in the pores ("suggesting"). The PXRD and DSC conditions are not described. P1·M1·M2·M3·M4·R6·C15·X13

Charge of the silica and of the drugs

The zeta potential of SBA-15 was measured with a Malvern Nano-Zetasizer at 37 °C from pH 1 to 10, with the pH set by concentrated HCl or NaOH. The percent ionized of each drug was calculated from the pKa values with the Henderson–Hasselbalch equation (Figure 1), and crystalline pH-solubility profiles were calculated the same way (Figure S4). P4·P10·P11

Adsorption isotherms

A small aliquot of methanolic drug stock was added to an aqueous medium containing SBA-15 in the amount that would be present after a dissolution test of the 20% formulation; methanol stayed below 0.5%. The suspensions were equilibrated for 2 h at 37 °C (preliminary tests showed fast equilibrium), centrifuged at 21,100 × g for 10 min and the supernatant analysed by HPLC. The media were 0.1 N HCl (pH 1), 50 mM acetate buffer (pH 4.5 and 5.5) and 50 mM phosphate buffer (pH 6.8), each containing 100 µg/mL HPMC (PharmCoat 606) to prevent crystallization. All points were measured in triplicate and plotted with standard deviations. The initial concentrations and sample volumes are not given. P2·R22·M5·M8·M9·M10

Release at a single pH and after a pH shift

The dose was set so that full release would give a concentration equal to the amorphous solubility of the free base: 50 µg/mL KET, 150 µg/mL CLZ and 75 µg/mL ATZ (values cited from the group's earlier work, determined by UV extinction). Each test used 50 mL at 37 °C, in triplicate. Single-pH tests ran for 60 min in each of the four media. Two-step tests ran 30 min in 0.1 N HCl, then concentrated NaOH raised the pH to 4.5, 5.5 or 6.8, and release was followed for another 90 min; a fed-state sequence started at pH 4.5 and shifted to 5.5. How samples were taken and separated in these closed tests is not described. P5·P6·M1·M2·M3

Competitor experiments

At pH 6.8, KET and CLZ adsorption on SBA-15 was measured with 50 mM TEA or 0.5 M urea added (Figure 6); the drug concentration used is not stated. In a second experiment, TEA (50 mM) was added at the moment of the pH shift in two-step tests from pH 1 to 6.8 (Figure 7). The final pH after the TEA addition is not reported. P3·P7·X7

Absorptive dissolution

KET-MPS was also tested in the group's flow-through absorptive apparatus (ref 35): a cellulosic hollow-fiber module with 100 cm² of membrane and 15 nm pores, 50 mL donor, pH 6.8 phosphate receiver, 4 mL/min flow on both sides, 37 °C. A 10 µm cannula filter kept silica particles out of the fibres. The formulation was held 30 min at pH 1, the pH was raised to 6.8 and absorption began; donor samples were analysed by HPLC after centrifugation and the receiver by an in-line UV probe, up to 270 min. P8

Experimental analysis

Result 1: silica turns negative, and two of the three drugs follow it

SBA-15 carried a small positive zeta potential at pH 1 (about +1.7 mV) and pH 2 (about +0.5 mV), was slightly negative at pH 3 (about −1.6 mV), fell steeply to about −22.2 mV at pH 6 and levelled at about −28.5 to −29.6 mV from pH 8 to 10 (Figure 2). The authors place the isoelectric point at pH 2, in line with the literature; on the plot the zero crossing lies between pH 2 and 3. R2·C5·X12

Table A puts the charge of each drug next to its adsorption and its release at the same pH. The adsorbed amounts are the highest point of each isotherm in Figure 3; because the concentration ranges differ between drugs, the comparison is meaningful within one drug across pH, and the fitted monolayer coverage from Table S1 is added as a second measure.

Table A. Charge, adsorption and release of the three drugs at four pH values. Ionization read from Figure 1, adsorption from Figure 3 and release from Figure 4 (all approximate); Q_m from Table S1. R1·R3·R4·R5·R7·R8·R9·R21

For KET and CLZ the pattern is clear. Adsorption rises as the silica becomes more negative, and release falls where adsorption is high: KET released about 92% and 87% at pH 1 and 4.5 but only about 60% and 55% at pH 5.5 and 6.8; CLZ released about 93–94% at pH 1 and 4.5, about 88.5% at pH 5.5 and about 73.5% at pH 6.8. CLZ did not adsorb at all at pH 1, which the authors attribute to repulsion between the cationic drug and the silica; the silica is only slightly positive there. Two details are easy to miss. First, CLZ is almost fully charged at pH 4.5 and 5.5, yet its adsorption still rises steeply to pH 6.8, so in this range the growing silica charge, not the drug charge, drives the change. Second, at low concentration the KET pH 6.8 isotherm lies below the pH 4.5 and 5.5 isotherms (about 25 mg/g at about 20 µg/mL against about 35–39 mg/g near 22 µg/mL) and only overtakes them at the highest concentrations. The authors describe the KET change from pH 4.5 to 6.8 as "a slight increase" and suggest that the loss of charged KET is balanced by more deprotonated silanols. R3·R4·R7·R8·C26·C27·C36·C37

ATZ behaves differently. It adsorbs about as much at pH 1, 4.5 and 5.5 (about 37.5–40.5 mg/g at the highest points) and about half as much at pH 6.8, and its monolayer coverage falls from 0.057–0.058 at pH 1 and 4.5 to 0.006 at pH 6.8. Its release is highest at pH 1 (about 70%, released at once and then flat) and low at every higher pH (about 33–40.5% at 60 min after a burst of about 14.5–21%). The authors attribute the relatively high adsorption at low pH "most likely" to ATZ's lower pKa, which changes the balance between adsorption and solvation, and the low release at higher pH to its low solubility, which makes the formulation more hydrophobic so that water enters the pores more slowly ("is expected"). R5·R9·R21·C40·C43

On the activity scale of Figure S5 (pH 6.8), CLZ reaches about 53 mg/g at an activity of only about 0.19, KET about 66 mg/g at about 0.445, and ATZ only about 21 mg/g at about 0.86; the authors rank the monolayer coverage at pH 6.8 as CLZ > KET > ATZ and read the rise of the CLZ coverage with pH as more adsorption sites from silica ionization ("suggests"). Note by the AI writer: the amorphous solubilities used for this activity scale are not stated; matching the plotted activities to the Figure 3 concentrations implies about 83 and 620 µg/mL for KET and CLZ, far above the free-base values of 50 and 150 µg/mL, so the scale appears to use pH 6.8 solubilities that include the ionized drug. For ATZ the free-base value of 75 µg/mL reproduces the plotted activity. The printed unit of Q_m, "mg/mol", is not an amount per mass of silica, so the Q_m values can only be compared with each other. R20·C42·X8·X9·E3

Result 2: after the pH shift, KET and CLZ lose about 30% of the dissolved drug; ATZ does not

This is the central observation of the paper (Figure 5). Before the shift, KET and CLZ were released almost completely in pH 1. At the first sample after the pH was raised, part of the dissolved drug was gone, and the level stayed nearly flat for the next 90 min. ATZ, by contrast, did not lose drug when the pH was raised to 5.5 or 6.8. Table B lists the values.

Table B. Two-step release, shift at 30 min, all values read from Figure 5 (approximate). For KET and CLZ the two points plotted at 30 min are clearly before and after the shift; for ATZ the figure does not mark their order. R10·R11·R12

The authors summarise the KET loss as "almost 30-35%" on the shift to pH 5.5 or 6.8 and the CLZ loss as "approximately 30%" for pH 1 to 6.8. Read from the plot, KET lost about 32 points on the shift to 6.8 but about 21 points at the shift to 5.5 (about 25 points by 120 min). Both drugs also lost about 10–15 points on the smaller shift from pH 1 to 4.5, which the paper does not discuss; ATZ lost about 10 points on that shift and then levelled at about 63% (the text: 60%). In the fed sequence KET ended at about 68.5% (the authors: 70% at pH 4.5 and no further release at 5.5), CLZ dropped by about 12 points to about 82.5% (the authors: a 15% reduction), and ATZ released poorly, ending at about 38.5%. R10·R11·R12·X11

Because a falling concentration after a pH shift usually signals precipitation, the authors collected the solid at the end of the pH 1 to 6.8 tests and examined it by PXRD (Figure S3). It showed only a broad amorphous halo, so they ruled out crystallization and proposed the alternative: the drug released in acid re-adsorbed onto the silica. R13·C17

Result 3: a competing cation removes the drugs from the silica and largely prevents the loss

Table C. Competitor experiments. Adsorption read from Figure 6, release from Figure 7 (approximate). R14·R15·R16

At pH 6.8, TEA lowered KET adsorption by about 87% and CLZ adsorption by about half, which the authors read as competitive adsorption of the cation onto the silica ("indicating"). Urea left CLZ unchanged and lowered KET from about 66 to about 47.7 mg/g, which the authors call "only small" and read as "some extent of hydrogen bonding" of uncharged KET in addition to electrostatic binding. Their conclusion for KET is that the TEA result "confirmed" primarily electrostatic binding to SiO⁻ groups; for CLZ, electrostatic interaction "seemed" to dominate. When TEA was added at the pH shift, the loss shrank: KET went from about 96% to about 93% and then drifted to about 88%, CLZ from about 96% to about 85% and then about 80%. The authors describe this as "consistent with less re-adsorption" and, in the discussion, as support for "the conjecture" that the concentration drop is re-adsorption; the abstract states that further investigations "demonstrated" it. The text also says that with TEA the KET concentration "did not decrease, plateauing at 95% release", which is more than Figure 7a shows. R14·R15·R16·C20·C21·C28·C29·C35·C38·X3·X5

Result 4: an absorptive sink frees little of the bound KET

If re-adsorption were an easily reversed equilibrium, removing dissolved drug through a membrane should pull drug back off the silica, as the authors had seen for ATZ-MPS (ref 28). For KET-MPS the effect was small (Figure 8). In this apparatus about 89–90% of the KET dose was in solution at pH 1. At the shift the donor concentration fell abruptly from about 44.8 to about 29.5 µg/mL and then declined steadily to about 5.9 µg/mL at 270 min, which the authors attribute to absorption across the membrane. The receiver concentration rose to a maximum of about 2.9 µg/mL between about 40 and 58 min and then fell to about 0.63 µg/mL, following the donor. Total release dropped to about 54–55.5% at 45–90 min and climbed only to about 69.5% at 270 min. The authors describe this as an increase "only slightly (by 10%)" over the 4 h of absorption and conclude that KET adsorption at high pH "seems" to come from a strong interaction: the results "suggest that adsorption was at least partially irreversible at high pH or that release was very slow". R17·R18·R23·C22·C23·C24·C32

Note by the AI writer (not discussed in the paper): the absorptive run starts lower than the closed two-step test of Figure 5a (about 89–90% against 96–98% in acid, about 54–60.5% against 64–69% after the shift), and the way total release was calculated from donor and receiver data is not described. The "10%" matches the rise from the first point after the shift (about 60.5%) to 270 min (about 69.5%); from the 45–90 min level the rise is about 15 points. X10

A mass-balance check (AI writer's arithmetic)

Note by the AI writer (not discussed in the paper): the isotherms used the same amount of silica as a dissolution test. With 20% w/w drug, the silica mass is four times the drug mass, so a 50 µg/mL KET dose comes with about 0.2 mg/mL SBA-15 and a 150 µg/mL CLZ dose with about 0.6 mg/mL. The loss after the shift to pH 6.8, about 32 points for KET (about 16 µg/mL) and about 30 points for CLZ (about 45 µg/mL), then corresponds to about 80 mg/g and about 75 mg/g of drug on the silica. The pH 6.8 isotherms at the concentrations left in solution (about 32 µg/mL for KET, about 103 µg/mL for CLZ) lie lower, below about 48 mg/g for KET and at about 44–53 mg/g for CLZ. The loss is therefore of the size that adsorption could explain, but somewhat larger than the isotherms predict at equilibrium; the drug released into solution in acid and the drug added from a methanol stock in the isotherm experiments may not meet the silica in the same way. The paper itself does not compare these numbers. R3·R4·R10·R11·P2·P5

Discussion

The authors' explanation: adsorption against solvation, switched by pH

The authors explain the results with a balance between two tendencies of a dissolved drug: to stay solvated in water, and to bind to the silica. Adsorption, they write, is "essentially a competition" between the drug–silica interaction and the solvation energy of the drug, and a higher solvation energy lowers the drive to adsorb (ref 45). pH moves both sides of the balance at once.

In the authors' words, the change in balance "leads to drug being re-adsorbed from the solution onto the silica surface upon pH-shift", and TEA support for "the conjecture" comes from Figure 7. For ATZ, the low ionization at intestinal pH "precludes electrostatic interactions" and "most likely" accounts for the absence of re-adsorption. Across the three drugs, they conclude that pKa "affects the solubility and ionization of drugs, which in turn governs the adsorption tendency and drug release". C31·C20·C41·C44

What the results imply for formulation

The authors draw three practical points. First, for weak bases with a high pKa, the effect of pH on the solution concentration must be tested: "Although dissolution under acidic conditions is expected to improve drug release", the concentration "may decline upon transfer to intestinal conditions due to re-adsorption", which is "ultimately expected to reduce the drug bioavailability". Second, weak bases formulated in MPS "would be expected to show good release in fasted state gastric conditions but may have reduced availability upon transit to intestinal conditions", and MPS "might not be the most suitable approach for weakly basic drugs with high pKa or a pKa above intestinal pH". Third, in vivo-relevant conditions should be built into in vitro testing, because the adsorption–dissolution balance in the gut will also depend on other species that adsorb competitively (other cations, biological surfactants) and on solubilizing bile salt micelles, which "are speculated" to matter. From the fed-state sequence they add that KET-MPS shows "no food effect" from a pH perspective. C34·C51·C52·C46·C47·C33

The AI writer's reading (not stated in the paper): the results give a simple flag for silica carriers: the fraction of a weak base that is still charged at pH 6–7. KET, only about one-third charged at pH 6.8, still re-adsorbed, so the threshold is not "mostly charged" but "appreciably charged". The medium contained 100 µg/mL HPMC throughout, so the loss is not prevented by a crystallization inhibitor; it is a surface effect. A competing cation kept the drug in solution in vitro, but whether a pharmaceutically acceptable cationic excipient, a surface-modified silica or a different carrier would do the same was not tested. For low-pKa bases such as ATZ the issue is different: there is little re-adsorption, but release at intestinal pH remains low. R1·R15·R16·M5

Where the conclusions apply

All data come from one silica (SBA-15 with 7.1 nm pores), one drug loading (20% w/w), doses equal to the free-base amorphous solubility, 50 mL of simple buffers containing HPMC and no bile salts, and 37 °C. Competitor experiments cover KET and CLZ, and the absorptive run covers KET only. No absorption data were collected for CLZ, and no in vivo data for any drug. The general statement in the conclusions, that all three drugs "primarily interacted with silica via electrostatic interactions" because they are weak bases, is broader than the data: it was not tested for ATZ, whose adsorption falls where it loses its charge. A1·A2·A3·A4·A5·C48·X4

Limitations

Limitations stated by the authors. The authors note that the impact of re-adsorption "needs to be evaluated in greater depth in an in vivo context", that gastrointestinal fluid components (competing cations, surfactants, bile salt micelles) will influence the balance, and that predicting in vivo performance from in vitro tests is "a significant challenge". They leave open whether KET adsorption at high pH is "at least partially irreversible" or whether release is "very slow". C39·C46·C47·C32

The AI writer's own judgement (1): how firmly re-adsorption is shown. The loss of dissolved drug after the shift is measured directly and is large, and the TEA experiments point the same way twice: TEA strips the drugs from silica and lessens the loss. Two gaps remain. PXRD of the collected solid excludes a crystalline precipitate only; an X-ray amorphous drug precipitate would also show no peaks, and the solid contains the silica. And TEA is a base added at 50 mM; the final pH after its addition is not reported, nor whether it changes drug solubility, and the drug concentration in the Figure 6 test is not given. The amount of drug on the silica after the shift was never measured directly. Taken together, re-adsorption is the most consistent explanation of the data, but the abstract's "demonstrated" is stronger than the discussion's "conjecture", and the latter fits the evidence better. X6·X7·C20

The AI writer's own judgement (2): the electrostatic mechanism and the adsorption–release link. For CLZ the competitor result is clean (TEA halves adsorption, urea has no effect). For KET, urea lowered adsorption by about 28%, more than "small", so hydrogen bonding is a real part of KET binding. The extension of the electrostatic mechanism to all three drugs does not fit ATZ, and the abstract's statement that "higher adsorption, in turn, led to a lower extent of drug release" does not hold for ATZ across pH: ATZ adsorbs most and releases most at pH 1. The authors explain ATZ separately, by solubility and formulation hydrophobicity, which is an explanation they did not test. X4·X5·X15·C43·C49

The AI writer's own judgement (3): text against figures. Several numbers in the text differ from the plots: the KET loss at pH 5.5 is about 21–25 points rather than 30–35; ATZ at pH 5.5 plateaus at about 33%, not 40%; KET with TEA declines to about 88% rather than plateauing at 95%; and a 10–15-point drop after the pH 1 to 4.5 shift is plotted but not discussed. Four figure references name the wrong panel (Figure 4a for ATZ, Figure 5b for KET, Figure 5c for CLZ, Figure 3a for CLZ), and the CLZ pKa is given as 7.5 in the introduction and 7.3 in Figure 1 and the results; the plotted ionization and solubility curves are consistent with 7.3. The no-food-effect statement rests on KET only; ATZ released about 38.5% in the fed sequence against about 63–75% in the fasted ones. None of these changes the main conclusions, but readers should take values from the figures. X1·X2·X3·X11·X14

Conclusion

The paper shows, for three weak bases on unmodified SBA-15, that pH acts on both partners of the drug–silica pair. The silica turns from near neutral at pH 1 to strongly negative at intestinal pH, and a weak base turns from fully charged to partly or wholly neutral. Where the drug stays appreciably charged at intestinal pH (KET, CLZ), it binds the silica: adsorption rises and release falls with pH, and drug that dissolved in acid partly leaves solution again after a pH shift, by about 30% for the pH 1 to 6.8 shift. A competing cation removes most of the bound drug and most of the loss, and an absorptive sink frees little of it over 4 h. Where the drug is uncharged at intestinal pH (ATZ), there is no re-adsorption, but release stays low for other reasons. The authors' practical message is that pKa matters for MPS formulations of weak bases, that pH-shift testing should be part of their evaluation, and that MPS "might not be the most suitable approach" for weak bases with a high pKa. R2·R3·R4·R5·R10·R11·R12·R14·R15·R16·R17·C44·C52