IVRT & IVPT Support

The first question in any in vitro permeation study is which skin to use. The industry answer is usually a convention: excised human skin if available, pig ear if not, rodent skin as a last resort. The convention is rarely tested against the compound in hand.

We do it the other way round: get an in vivo reference first, then screen the models against it.

Epidermal residual fraction measured on living skin after application: 72.09%. That is the yardstick. Ten candidate models were then run under one protocol—five porcine skins (Bama miniature pig, one-month-old piglet, inner ear, outer ear, abdominal), three rodent skins (mouse abdomen, nude mouse, rat abdomen and back) and excised human skin.

The result was counter-intuitive. The closest match to the living value was porcine outer ear skin at 72.97%—within 0.9 of a point. Excised human skin came in at 87.57%, fifteen points off. Harvesting, freezing and thawing had changed it; it was no longer the living tissue it is assumed to stand in for.

Model	Residual	Distance from in vivo
In vivo reference	72.09%	—
Porcine outer ear	72.97%	+0.9
Excised human skin	87.57%	+15.5

For porcine outer ear the three endpoints closed: epidermal residual 72.97% + intradermal retention 20.96% + permeated 4.55% = mass balance 98.48%.

Two conclusions. “Pig skin” is not one model: inner and outer ear of the same animal differ; one-month-old and adult differ; ear and abdomen differ. And the model is compound-specific—this compound’s answer does not transfer to the next. What transfers is the method: get an in vivo anchor, run the candidates, let the data choose.

The analytics were built alongside: specificity established separately for each of three matrices; linearity 0.4395–8.79 µg/mL; LOD 7.34 ng/mL, LOQ 14.68 ng/mL; recovery 95–105%; 24 h stability deviation <2%. The method was then run on originator, in-house and generic creams side by side. The in-house formulation closed at 95.18% mass balance with a lower residual than the originator—a formulation that needed further work, and a method with the resolution to show it.

For the separate formulation decision, see the later trial-cream skin-distribution comparison. The texture-validation and skin-comparison batches were not the same.

Of the three endpoints, intradermal retention is the one most easily underestimated. The usual procedure—wash off the surface, drop the skin in a vial, add solvent, agitate, inject—looks fine until you run a recovery.

We compared two extraction approaches on the same skin, the same compound, the same chromatography.

Extraction	Recovery	RSD
Direct solvent agitation	59.76% (58.82%, 60.70%)	—
Liquid-nitrogen cryo-milling → ultrasonic cell disruption → extraction, low	97.60%	2.71%
mid	99.15%	1.20%
high	100.72%	0.58%
mean	99.16%	—

Direct agitation missed four-tenths of the drug—not as random scatter but as a systematic low bias. Without a recovery study you would carry a stable, reproducible number at sixty percent of the true value into the report.

Why it misses. After 24 h the drug has reached the deep dermis, where solvent does not soak. And the receptor fluid contains ethanol—needed for sink conditions with a lipophilic compound—which over 24 h dehydrates and stiffens the tissue. Agitation does almost nothing to a sample like that.

Our procedure. Skin cut into fine strips → milling tube with two steel beads → liquid nitrogen 2–3 min → tissue mill, 60 Hz, 30 s → re-freeze, mill again → powder to vial, 2 mL water rinse → ultrasonic cell disruption (200–400 Hz, 5 s on / 5 s off, 20 cycles) → 18 mL acetonitrile, 15 min agitation + 15 min sonication → evaporate at 60 °C → reconstitute in 2 mL 90% methanol. The double freeze-mill is the point: one pass leaves strips, the second makes powder.

A knock-on effect. Excised human and rodent skin, after 24 h in ethanol-containing receptor fluid, become dry and hard; even after cryo-milling most of the tissue stays in strips and the mass balance will not close. Porcine ear skin under the same conditions retains enough moisture to powder in two passes. So “which skin” is also a question of whether you can get the drug back out at the end. Both belong in the design stage.

Most studies report one number: the amount permeated. We report three and require them to sum to roughly 100%—surface residual + intradermal retention + permeated = mass balance. Not for an extra metric, but because with one number you cannot know what you have lost. Drug can stick to the cell wall, evaporate during weighing, resist extraction, or leave with the wash—each lowers the permeated fraction, and the permeated fraction alone will never tell you which happened.

Here is one real diagnostic.

	Pilot	Formal run
n	4	5
Diffusion cell	1.5 cm diameter	0.95 cm² effective area
Applicator	flat head of a drawing pin	200 µL pipette tip, end heat-sealed and ground flat
Weighing	recorded immediately	recorded once reading settled
Surface residual	80.26 ± 10.53%	87.57 ± 4.39%
Mass balance	87.09 ± 8.52%	101.10 ± 7.14%
Distance from 100	−13	+1

Where the thirteen points went. Two causes, neither in the chromatography. The applicator was too large for the cell: cream touched the upper chamber wall and that fraction was neither on the skin nor in the wash. And the room was warm: water evaporated from the cream on the balance, the reading drifted, and the recorded dose was not the real dose. Change the applicator, change the cell, wait for the balance—and the same skin closes at 101%. The same method on porcine outer ear: 98.48%.

The point is not that 87 became 101. It is this: had only permeation been measured, the pilot data would have looked entirely normal—stable, reproducible, no warning sign. The wall-contact and weighing-drift problems would have stayed in the method, into the formal study, into the report. Mass balance is the only mechanism by which a method speaks up before the report is written.

The in vivo reference followed the same logic. First version, cream applied directly and unprotected: 55.14 ± 12.23%, wide scatter, wash fluid leaking from a hand-held cap. Second version, cap fixed with adhesive and sealed for 24 h: 72.09%, scatter markedly narrower. The first number a method gives is usually the operator’s number, not the method’s. Variance and mass balance are how you tell them apart.

A product comparison begins with a defined question. For a locally acting active, the amount retained in skin can be more informative for delivery than the amount passing completely through it. Receptor-fluid suitability, skin model, dose and analytical recovery still need to be set for the active and endpoint in question.

In one ex vivo comparison, three marketed ointments containing the same active were evaluated using porcine ear skin, a 25% ethanol receptor fluid and a 20-hour exposure. Skin retention was measured in six cells per product, with each cell's applied dose recorded separately.

	Retention	RSD	vs orig.
Originator	18.0 ± 0.3%	1.8%	—
Generic A	12.5 ± 1.0%	8.1%	p < 0.001
Generic B	12.0 ± 0.6%	5.0%	p < 0.001

What this comparison supports. Under the stated laboratory conditions, the originator showed higher skin retention of the measured active than either comparator. The values do not establish which product works better in people, nor do they by themselves establish clinical or regulatory equivalence.

What carries over to another product question. Define a shared, quantifiable target and an endpoint relevant to the intended claim; then agree the skin model, dosing, receptor system, analytical method and quality checks before comparing products. Different actives or different claimed benefits may require a different study rather than an IVPT ranking.

Have two topical products to compare? Start with a product-comparison feasibility enquiry. Laboratory execution and pricing are scoped per project.

One laboratory, one skin type, one in-house method, one report—that is the norm. Nobody knows whether a second laboratory would get the same number, so nobody knows whether the number can be trusted.

We took part in a comparison across three laboratories, two countries and two species: one cosmetic active, one harmonised protocol, each laboratory running it independently. One used full-thickness human skin, two used full-thickness pig skin. Vertical diffusion cells, a 0.3% emulsion, 5 mg/cm², 16 h. Compartmentalised endpoints: surface wash, stratum corneum by tape stripping, epidermis + dermis, receptor fluid.

Validate first, then compare. Each laboratory established linearity, reproducibility, LOD/LOQ and carryover; matrix effect and extraction recovery were determined separately for each skin compartment—four matrices, four sets of correction factors. Mass balance 100 ± 15%; cells outside it excluded.

	Human	Pig, lab B	Pig, lab C
Surface wash	82.3 ± 8.4	74.7 ± 4.8	70.9 ± 8.4
Stratum corneum	1.32 ± 0.48	0.37 ± 0.22	0.7 ± 0.7
Epidermis + dermis	1.51 ± 0.51 (Ep) + 10.0 ± 4.3 (D)	7.66 ± 3.84	14.3 ± 2.8
Receptor fluid	1.78 ± 0.93	3.46 ± 2.05	1.5 ± 0.4
Mass balance	96.9 ± 6.8	86.7 ± 4.9	87.3 ± 5.6

After 16 h, 70.9–82.3% of the active remained unabsorbed on the surface; skin bioavailability was 11.1–15.8%; the absorbed fraction sat mainly within the skin. The largest inter-laboratory difference was 1.9-fold, in line with the 1.8-fold variability reported in the literature.

The decisive line is not in the table but between the columns: the two pig-skin laboratories differed from each other by more than pig differed from human. No species effect. Pig skin can stand in for human skin—provided the method has been validated and corrected in each laboratory.

Having used human skin is not, by itself, a guarantee the number is sound; reproducibility across laboratories is. Skip the compartment-specific matrix effect and recovery, and two laboratories’ numbers cannot be placed side by side.

Published: J Chromatogr B 2018;1093–1094:77–79, doi:10.1016/j.jchromb.2018.06.048.