Making a peptide and its permeation enhancer dissolve together: dispersion in a gel-forming polymer

Background and motivation

Oral peptide drugs have returned to the market: oral semaglutide (Rybelsus, 2019), octreotide (Mycapssa, 2020) and voclosporin (Lupkynis, 2021) are all built on enabling formulations (Introduction). Semaglutide and octreotide are large and hydrophilic, so they cross the gut wall poorly and are formulated with a permeation enhancer (PE): the semaglutide tablet contains salcaprozate sodium (SNAC), and the octreotide capsule is enteric coated and contains sodium caprylate (C8) together with surfactants and lipids. Voclosporin, a modified cyclosporine, is instead limited by solubility and is given as a self-emulsifying drug delivery system (SEDDS). C7 The outcome differs accordingly: the authors cite oral bioavailability of 0.4–1% for semaglutide with SNAC, 0.7% for octreotide and about 50% for voclosporin. R26 Even with an enhancer, then, peptide absorption remains very low.

Permeation enhancers are a chemically diverse group. Their proposed mechanisms include fluidizing the membrane and opening tight junctions, and an ideal enhancer acts quickly and reversibly so that the protective barrier of the gut recovers. C8 Two requirements recur in the literature the authors summarize: the enhancer must reach a high local concentration, and it must be co-localized with the peptide at the absorption site, because there is only a short window for peptide absorption after the enhancer has acted on the membrane. C8 For sodium decanoate (SD, also called sodium caprate or C10), studies suggest that at least 10 mM is needed at the membrane surface, and SD itself is absorbed within 15–30 min. R28·C36 This is hard to arrange with a solid oral dosage form, especially one aimed at the small intestine, where dilution, motility and the different solubilities of peptide and enhancer can pull them apart. C10

The authors build on two lines of earlier work. The first is a pharmacokinetic precedent: Baluom et al. (ref 45) placed the poorly absorbed small molecule sulpiride together with SD in HPMC matrices of different grades so that drug and enhancer released together over different periods; after intra-intestinal dosing in rats, synchronous release gave better absorption than non-synchronous release, and synchronous release over 1 h was better than slower synchronous release. R27 The second is from the authors' own field of amorphous solid dispersions (ASDs): with copovidone, a drug can be released "congruently" with the polymer, at about the rate of the neat polymer, because the polymer controls the release (refs 46–51, including Indulkar 2019 and Saboo 2020). C9

The question the paper asks follows directly: if PVPVA can make a lipophilic small molecule leave together with the polymer, can it also make a peptide and a permeation enhancer leave together? The authors chose octreotide acetate (OCA) as a model hydrophilic peptide and cyclosporine (CYC) as a model lipophilic peptide, SD and SNAC as enhancers "of contemporary interest", prepared binary and ternary dispersions by solvent evaporation, checked their solid state by powder X-ray diffraction (PXRD), and measured the surface-normalized release of every component. A1·A2·P1·P3·P4

Background concepts

Surface-normalized (intrinsic) dissolution

An intrinsic dissolution rate (IDR) is measured from a compact of constant exposed area, so that the rate reflects the material and not changes in particle surface. Here the authors pressed 100 mg of material into the die of a Wood's apparatus with a 0.5 cm² face, rotated it at 100 rpm in 100 mL of 50 mM pH 6.8 phosphate buffer at 37 °C, and sampled every 5 min for 30 min and then every 10 min to 60 min. P4 Each component was assayed separately: OCA and SNAC by reversed-phase HPLC, CYC by HPLC with the column at 75 °C, and PVPVA by aqueous size-exclusion chromatography with UV detection at 205 nm. SD could not be assayed. P2·R6 The rate of each component is reported per unit area and per unit of its own weight fraction (Eq. 1):

In plain terms: $k$ is the slope of the linear part of the concentration–time curve, $V$ the medium volume (100 mL), $S$ the exposed die area (0.5 cm²) and $x$ the weight fraction of the component in the compact; dividing by $x$ makes the rate of a minor component comparable with that of a major one. E1 If all components of a dispersion leave in the same proportion as they are present, they give the same $R$. That is the paper's working definition of synchronous, or congruent, release. The slopes $k$ are not printed; the rates are shown only as bar charts (Figures 2 and 3). R2

Congruent release and the gel layer

When a PVPVA dispersion meets water, its surface hydrates into a swollen, viscous gel layer; the authors cite earlier work showing this layer can exceed 200 µm. R29 If the drug and any other ingredient are held in that layer, they can only leave as fast as the layer erodes, and the polymer then sets the rate of everything. This idea underlies the paper's interpretation, but the gel layer itself was not observed or measured in this study. C45·X6

The two enhancers and their self-association

SD is the sodium salt of a C10 fatty acid; at pH 6.8 it forms vesicles above a critical vesicle concentration (CVC), and at pH 1.6 it is expected to convert to the poorly soluble decanoic acid (DA). C26·C33 SNAC is an N-acylated amino acid salt that is intrinsically fluorescent (emission at 414 nm) and, as this paper shows, self-associates above about 0.1 mg/mL, which quenches its own fluorescence. R19·R24·C34 Octreotide is also autofluorescent, with an emission maximum around 350 nm when excited at 296 nm, which the authors use to look for interactions in solution. R19·P7

Drug loading

Drug loading (DL) here means the weight percentage of peptide in the dispersion. In all ternary dispersions the enhancer:PVPVA ratio was fixed at 50:50 w/w, so a 1.6% DL ternary is 1.6:49.2:49.2 peptide:enhancer:polymer, and a 20% DL ternary is 20:40:40. P1 A binary peptide:PVPVA dispersion at the same DL therefore contains about twice as much polymer as the ternary. X7

Methods in detail

The study is a single experimental programme with three parts: making and characterizing the dispersions, measuring the release of each component under one set of conditions, and a group of supporting solution and solid-state measurements meant to explain the release behaviour. No model was fitted.

Part 1: dispersions and their solid state

All dispersions were made the same way: the components were dissolved in methanol in a round-bottom flask, the solvent was removed on a rotary evaporator with the water bath at 45 °C, and the product was kept under vacuum overnight, ground with mortar and pestle and stored in a desiccator. P1 The authors list SD:PVPVA, SNAC:PVPVA and DA:PVPVA binaries "at various w/w ratios", OCA:SD:PVPVA and OCA:SNAC:PVPVA at 1.6% DL, and CYC:SD:PVPVA and CYC:SNAC:PVPVA at 1.6, 5, 10 and 20% DL, always with enhancer:PVPVA at 50:50. P1 Binary CYC:PVPVA dispersions at the same four loadings and a neat amorphous CYC also appear in the results, but their preparation is not described, and no data for the DA:PVPVA dispersions are shown. X14

The materials as received differed in solid state. Neat OCA and PVPVA gave only an amorphous halo, while SD, SNAC and the as-received CYC were crystalline (Figure 6). R17 This matters for the interpretation: in the dispersions the peptide could be amorphous while the enhancer stays crystalline, so congruent release, if found, cannot simply be put down to a single homogeneous amorphous phase.

Part 2: release of every component

The central measurement is the surface-normalized dissolution described above, run in triplicate on neat OCA, SNAC and PVPVA, on SNAC:PVPVA and SD:PVPVA binaries, on both OCA ternaries at 1.6% DL and on the full CYC series (Figures 2–4). P4·R1·R2·R10 Two features of the design deserve attention. First, because every component except SD was assayed in the same medium, the paper can compare release profiles directly rather than infer one from another. P2 Second, SD could not be measured "due to analytical constraints"; for neat SD the authors observed that the compact had dissolved completely within 7–9 min. R6

A limited test at gastric pH followed: the 1.6% and 20% DL CYC:SD:PVPVA ternaries were run in pH 1.6 medium (NaCl and HCl), and for the 20% DL ternary a two-stage run added 15 mL of concentrated pH 7.3 phosphate buffer after one hour to reach pH 6.8, and continued for another hour (Figure 5). P5·R16

Part 3: what the solution and the solid do

Four supporting measurements were intended to explain the release. Dynamic light scattering (DLS) of filtered and unfiltered dissolution samples looked for colloids. P6 Fluorescence spectra of OCA and SNAC, of mixed reference solutions and of dissolution samples looked for interactions between the components. P7 The pyrene I1/I3 ratio (SD) and SNAC's own emission at 414 nm located the onset of enhancer self-association, each with and without 1 mg/mL PVPVA. P8·P9 Dynamic vapor sorption (DVS) at 37 °C from 0 to 95% RH, in 5% steps, measured the water uptake of the enhancers, PVPVA and the two 50:50 enhancer:PVPVA dispersions. P10

Experimental analysis

1. On their own, the components dissolve at different speeds

From neat compacts, OCA released fastest: about 60% at 30 min and complete at 50 min, with a normalized rate of 4.5 mg min⁻¹ cm⁻². Neat PVPVA reached about 95.5% at 60 min (about 3 mg min⁻¹ cm⁻²), and neat SNAC only about 58.5% (about 2 mg min⁻¹ cm⁻²) (Figure 2a, b). R1·R2.a–c The authors read OCA's fast release as "presumably driven by its high solubility and amorphous nature" and state that OCA dissolves about twice as fast as SNAC. C11·C2 From the bars the ratio is closer to 2.5 (about 4.5 against 1.8), and OCA carries a large error bar of about ±1.05. X18 Neat amorphous cyclosporine was at the other extreme: less than 0.1% released in 6 h (Figure 4m). R14·C1

For an immediate-release product this means that peptide and enhancer would not arrive together: OCA would lead SNAC, while SNAC would lead CYC. The authors call such differences "likely to impact the effectiveness of the PE", and they single out a peptide that lags behind its enhancer as "likely to be an issue". C39

2. In PVPVA, peptide, enhancer and polymer release together

The SNAC:PVPVA 50:50 binary released both components at the same normalized rate, about 7.9 (SNAC) and 8.7 (PVPVA) mg min⁻¹ cm⁻², and both were complete within 25–30 min (Figure 2c, d). R2.d–e·R3·C13 Relative to the neat materials this is about a 4.4-fold increase for SNAC and 2.7-fold for PVPVA; the authors write "approximately fourfold" and "approximately doubled". C13·X18 Across ratios, the text reports that a 10:90 SNAC:PVPVA dispersion released each component slightly more slowly (complete within 40 min), while 20:80 and SNAC-richer dispersions released both congruently within 30 min (Figure S1, not available for this reading). R4·C14

Adding 1.6% OCA kept all three components in step (Figure 2e, f). Table A compares the rates.

Table A. Normalized release rates (mg min⁻¹ cm⁻²; mean of n = 3) in 50 mM pH 6.8 phosphate buffer at 37 °C, read from the bars of Figures 2b, 2d, 2f, 3b and 3d (approximate); the text gives 4.5, ~3 and ~2 for the neat materials, "around 8" for the SNAC ternary and "around 25" for the SD ternary. R2·R6

The profiles tell the same story as the bars: at 10 min OCA, SNAC and PVPVA stood at about 51, 47 and 47.5%, at 20 min at about 91, 86 and 92%, and all were complete by 25–30 min. R5 The authors conclude that "the release of the peptide and the PE could be controlled in the ternary system to be simultaneous", and in the discussion that the dispersion "eliminates the anticipated differences in release rates". C3·C40 Note that the common rate is not the rate of any neat component: the peptide, the enhancer and the polymer all leave faster from the ternary than from their own neat compacts. C15 OCA plateaus at about 105–107% of its nominal content, which suggests the nominal composition slightly underestimates the actual OCA content; the paper does not report an assay of the compacts. R5·X9

3. SD makes the polymer dissolve within about ten minutes

Neat PVPVA needed about 60 min for complete release; in the SD:PVPVA 50:50 binary, the polymer was about 46.5% released at 5 min and complete at 10 min (Figure 3a), a result the authors describe as "somewhat surprising". R1·R7·C17 The text reports that 20:80 SD:PVPVA was already enough to bring polymer release down to about 10 min, and that even 10:90 released the polymer within 25 min, about twice as fast as the neat polymer (Figure S2); "These results suggest that the amount of SD in the formulation could be varied to control the polymer release rate". R8·C18

In OCA:SD:PVPVA at 1.6% DL, OCA and PVPVA were about 37.5 and 38.5% released at 4 min, about 77 and 84.5% at 7 min, and complete at 10 min, with normalized rates of about 23 and 25 mg min⁻¹ cm⁻² (Figure 3c, d; text "around 25"). R9·R2.j–k The hydrophilic peptide therefore left about five times faster than from its own neat compact. C20 The authors call this "a highly synergistic improvement in the release of each component". C20 SD itself is described as released "presumably" at the same rate: an inference from the peptide and polymer data, since SD was not assayed. C19·X1 With the compacts nearly exhausted by the second or third sample, these high rates rest on few points, and the sampling times (about 4, 7 and 10 min) differ from the 5-min schedule stated in the methods. X8

4. Cyclosporine: synchronous only at very low loading

The cyclosporine series (Figure 4) is the paper's test of a lipophilic peptide and of loading. Table B collects the amount of CYC released after 60 min.

Table B. Percent of cyclosporine released at 60 min (other components in brackets), pH 6.8, 37 °C, n = 3; read from Figure 4 (approximate). Ternaries at enhancer:PVPVA 50:50. R10·R11·R12·R13

At 1.6% DL CYC released "congruently" with PVPVA from the binary over about 50 min; with SD the time for complete release fell to about 10 min, and with SNAC release took around 60 min, "similar to the binary system". C21·R10 The authors describe the SNAC ternary components as releasing "at the same normalized release rate"; no normalized rate is given for any CYC dispersion, and the profiles show CYC slightly behind its partners (about 75.5% against 85% SNAC and 89.5% PVPVA at 30 min). C21·X2

From 5% DL the picture changes. The binary nearly stopped releasing, a "large deterioration" that was "mitigated" by SD and "to some extent" by SNAC. C22·R11 At 10 and 20% DL the SD ternaries released the polymer completely within 40 min while CYC plateaued at about 40–54.5% (10% DL) and about 16% (20% DL): in the authors' words, "the release rate of PVPVA diverged from that of the drug". C23·R12·R13 The SNAC ternaries slowed every component, with relative rates that, "while not identical", stayed closer together than with SD. C24 The solid left after 60 min was amorphous by PXRD (Figure S3), so the plateaus were not caused by crystallization of the peptide. R15·C25 One comparison caveat: because the ternaries hold enhancer and PVPVA at 50:50, a binary and a ternary at the same DL also differ in polymer content (at 20% DL, 80% against 40% PVPVA). X7

5. Gastric pH

At pH 1.6 the 1.6% DL CYC:SD:PVPVA still released all its CYC, more slowly than at pH 6.8 (about 32% at 5 min, about 100% by 30 min), "despite the low solubility of SD at this pH". R16·C27 The 20% DL ternary reached about 12–14% (text: "about 10%"); after the shift to pH 6.8 it rose to about 19.5–23%, "a small increase", and the overall extent was similar to the single-stage run at pH 6.8 (about 16% at 60 min). R16·R13·C27 Both 20% DL series in Figure 5 carry the same legend label; the series that continues beyond 60 min has been taken as the two-stage run. X12

6. Solid state, colloids and interactions

PXRD (Figure 6) showed OCA and CYC amorphous in every dispersion. CYC:PVPVA 20% DL was fully amorphous, while all the enhancer-containing dispersions showed enhancer peaks: SD peaks "appeared largely unchanged", SNAC peaks were broadened, which the authors read as reduced crystallinity. R17·C28 In the CYC:SNAC:PVPVA patterns the strongest neat-SNAC reflection near 7.3° 2θ is not visible, whereas OCA:SNAC:PVPVA retains it; a different SNAC solid form would also fit those patterns. X15 The authors find it notable that PVPVA synchronized release "even though the PE was in crystalline form in the dispersion". C44

DLS of filtered dissolution samples of both OCA ternaries found colloids of about 10 ± 5 nm throughout the hour (Figure 7); unfiltered samples gave poor data with several peaks, which for SD is "consistent with" earlier reports of heterogeneous multilamellar vesicles. R18·C29 The SNAC samples also show a small population at about 100–1000 nm, and the caption does not say whether the plotted distributions are of filtered samples. X17

The fluorescence experiments (Figure 8) give a clean contrast between the two enhancers. OCA at 25 µg/mL gave the same spectrum (maximum about 33 near 350 nm) alone, with SD and PVPVA at 740 µg/mL, or with PVPVA at 25 µg/mL, which "likely suggests an absence of interactions" between OCA and SD or PVPVA. R20·C30 With SNAC, the OCA band shrank to a shoulder at equal concentrations and was not seen at all in the OCA:SNAC:PVPVA dissolution samples; the authors describe OCA fluorescence as "notably" and "highly quenched". R20·R22·C31 SNAC's own emission was also weaker at high concentration (about 6 at 740 µg/mL in the reference solution, against about 155 at 100 µg/mL). R19·R22·C32 The captions of Figure 8 do not fully match the legends (expected sample concentrations, sampling times and the composition of the reference in panel e), so these comparisons are best read qualitatively. X10

The self-association experiments explain part of this. The pyrene ratio of SD fell from about 1.9 to 1.0 between about 2 and 6 mg/mL, giving a CVC of 2.7 mg/mL without and 2.9 mg/mL with 1 mg/mL PVPVA, which the authors read as no notable change (Figure 9). R23·C33 Above the transition the ratio stayed higher with PVPVA (about 1.31 against 1.01 at 6 mg/mL), a difference the text does not discuss. X16 SNAC emission rose to a maximum near 0.075–0.1 mg/mL and fell to about 2.5–3 above 1 mg/mL, with or without PVPVA (Figure 10), "suggesting quenching due to self-association". R24·C34 Note by the AI writer (not discussed in the paper): the expected SNAC concentration in the dissolution samples, 490 µg/mL according to the Figure 8 caption, is well above this onset, so SNAC was largely self-associated when it quenched OCA. R22·R24

7. Water sorption

At 95% RH and 37 °C, SD:PVPVA sorbed about 90% water and SNAC:PVPVA about 86%, against about 37% for PVPVA, 22% for SNAC and 10% for SD (Figure 11). R25·C35 The authors state that the blends therefore sorbed more than predicted from their components. Weighting the neat values half and half gives about 23.5% and 29.5% (AI arithmetic; the paper does not print the prediction). C35·X5 The course of the isotherms differs between the blends: SNAC:PVPVA exceeds PVPVA from about 50–60% RH (about 55% against 24.5% at 90% RH), whereas SD:PVPVA tracks PVPVA up to 90% RH (about 27% against 24.5%) and jumps only at the 95% RH step. R25·X5

Discussion

How the authors explain synchronous release

The authors' explanation runs from the problem to a proposed mechanism. The chain below follows their Discussion; the steps marked "likely" are their interpretation, not measurements made in this paper.

On the speed-up, the authors write that "the mechanism of the enhanced release rate of the PEs in the presence of PVPVA (and vice versa) requires further investigation, but is likely due to enhanced water sorption in the binary mixture" (Figure 11), which "would increase the rate of gel erosion, and hence the rate of release of all of the components". C48 They regard the enhancement as "confirmed for PVPVA and SNAC dispersions (Fig. 2b and d)". C49

Note by the AI writer (not discussed in the paper): the sorption argument fits SNAC:PVPVA, whose isotherm departs from that of PVPVA from about 50–60% RH, better than SD:PVPVA, which tracks PVPVA up to 90% RH and departs only at 95% RH, although SD produces the larger speed-up. Vapour sorption at equilibrium is also not the same quantity as water uptake and erosion of a compact immersed in buffer. The sorption data are therefore consistent with the explanation but do not single it out. X5·R25

Why the release rate matters, as the authors place it in the literature

The Discussion places the result against what is known about enhancer delivery. Studies linking enhancer concentration to absorption were run under controlled conditions (rodent perfusion with known concentrations, or direct intra-intestinal dosing), whereas after oral dosing the fate of the dosage form is much less controlled. C10 Earlier work by Rubinstein and co-workers showed better absorption in rats from eroding tablets in which sulpiride and SD released together than from tablets in which SD released faster, and faster synchronized release beat slower. R34 (The paper cites this study as ref 77, which in the reference list is a review on colon targeting; the sulpiride study cited in the Introduction is ref 45. X13) Studies of eroding C10 and SNAC tablets for a GIP/GLP-1 peptide in monkeys, and of enteric-coated particulates and immediate- versus controlled-release tablets for a GLP-1 agonist, gave mixed signals on release speed. R31·R32 The authors conclude that the optimal release rate "is not clear, and is likely to depend on both the site of delivery, as well as the peptide and PE utilized", and that target release profiles "remain to be elucidated". C37

Against this background they highlight one practical point: Maher and Brayden postulated that a solid dosage form dissolving within 5–15 min could give the same bolus effect as a liquid-filled capsule, and the SD:PVPVA dispersion "readily" achieves complete dissolution from the IDR apparatus within 10 min "despite the limited surface area exposed". R30·C50 The dispersion approach thus "provides the advantages of a solid dosage form, coupled an extremely rapid and synchronous solubilization process". C51

Scope of the claim

The authors limit the gel-erosion mechanism to "low peptide loadings", and they note that a highly potent peptide would typically be present at a very low weight percent, where "little difference is seen in release rate for OCA versus CYC, eliminating the impact of peptide solubility". C45·C41 This holds for the SD ternaries, where both peptides released completely in about 10 min, but less so for the SNAC ternaries, where OCA was complete by 25–30 min and CYC reached about 95% only at 60 min. R9·R10·R5·X3 The extension to "different types of PEs" rests, for SD, on peptide and polymer data, since SD release was not measured. C43·X1 For the future, the authors suggest that other gel-forming polymers, for example PVP grades with different erosion rates, "may also synchronize release" and could give faster or slower release as needed. C47

What it suggests for formulation

Stated by the authors: a gel-forming polymer "may help synchronize the release of a hydrophilic peptide and a PE, which in turn may improve co-localization at the epithelial membrane"; the PE:polymer ratio can tailor the release rate "to some extent"; and other gel-forming polymers open routes to faster or slower release. C6·C46·C47

Note by the AI writer (not discussed in the paper): two practical consequences follow from the data. First, in this system the enhancer is not only the absorption aid but also the main lever on how fast the polymer dissolves (about 60 to 10 min with SD), so changing the enhancer dose changes the release clock of the whole dispersion. R1·R7·R8 Second, keeping a lipophilic peptide in step with the polymer required a very low peptide fraction: 1.6% worked and 5% already failed in the binary, so the approach suits highly potent peptides best. R10·R11

Limitations

Limitations stated by the authors. The authors state that the mechanism of the mutual release enhancement "requires further investigation". C48 They state that the optimal release rate for absorption is not known and target release profiles "remain to be elucidated". C37 They note that SD release could not be measured because of analytical constraints, so its synchronous release is described only as "presumably" occurring. R6·C19 They also note that at pH 1.6 SD "may not fully dissolve". C26

The AI writer's own judgement.

• The key link to absorption is untested. The paper measures in vitro release from a constant-area compact; it contains no permeability, cell, tissue or in vivo data, so the hoped-for improvement in co-localization at the membrane ("may improve") is a hypothesis. C6
• SD was never assayed. Synchronous release of SD, and the generalization to "different types of PEs", rest on peptide and polymer data only. X1
• The proposed mechanisms were not examined. No gel layer was imaged and no erosion rate measured; the > 200 µm gel thickness is cited from small-molecule dispersions, and the sorption link is a correlation that, for SD, rests on the 95% RH point. X6·X5
• Cyclosporine rates were not calculated. No normalized rate is given for any CYC dispersion; at 1.6% DL with SNAC, CYC lags slightly, and the claim that OCA and CYC behave alike holds for SD but not for SNAC. X2·X3 The description of the higher-loading SNAC ternaries as "still rapid, albeit incomplete" covers 60-min values of about 29.5% (10% DL) and 17% (20% DL). X4
• Few points behind the fastest rates. The SD-system rates of about 23–25 mg min⁻¹ cm⁻² rest on two or three samples before the compacts were exhausted, and sampling times differ from the stated schedule; several plateaus exceed 100% of nominal without a compact assay. X8·X9
• Confounded comparisons. Binary and ternary CYC dispersions at the same loading differ in polymer content as well as in enhancer. X7
• Reporting gaps. The Supplementary Information was not available for this reading; the Figure 5 legend and the Figure 8 captions disagree with the plotted series; the preparation of the binary CYC:PVPVA dispersions and of neat amorphous CYC is not described; and one cited study (ref 77) does not match the description given in the text. X12·X10·X14·X13
• Single conditions. One octreotide loading (1.6% DL), one enhancer:polymer ratio in the ternaries, one polymer and one release geometry were tested.

Conclusion

The study shows, with direct in vitro release data on every assayable component, that dispersing octreotide and SNAC in copovidone removes the difference in their intrinsic dissolution rates: all three components leave a compact at about 8 mg min⁻¹ cm⁻², faster than any of them alone. R2·R5·C3 Both enhancers accelerate PVPVA release, SD so strongly that the whole dispersion dissolves in about 10 min, and the authors suggest that the enhancer:polymer ratio can tune this. R7·R9·C17·C18·C46 The approach worked for the lipophilic cyclosporine only at 1.6% loading; at 5–20% the peptide decoupled from the polymer, in different ways with SD and SNAC. R10·R11·R12·R13·C5 The authors attribute synchronous release "likely" to erosion of the PVPVA gel layer and the speed-up "likely" to the higher water sorption of enhancer–polymer blends; neither mechanism, nor the hoped-for benefit at the epithelial membrane, was measured here. C45·C48·C6

For formulation, the paper turns a pharmacokinetic idea, that peptide and enhancer should arrive together, into a measurable dosage-form property: equal normalized release rates from one compact. The next steps the data point to are measuring the enhancer itself, imaging the gel layer and its erosion, testing other loadings, ratios and gel-forming polymers, and finally asking whether synchronous release in vitro improves permeation and absorption.