Endogenous Ascorbate Synthesis Confounds Oral Bioavailability Testing of Liposomal Vitamin C in Sprague Dawley Rats: Evidence from a Four-Arm Pharmacokinetic Study Including an Untreated Control

1. Introduction

Liposomal encapsulation is the dominant premium positioning for oral vitamin C, and the commercial argument for it rests almost wholly on bioavailability. The supporting human evidence is real but uneven. A scoping review of crossover trials comparing liposomal with non-liposomal ascorbate found greater uptake in nine of ten studies, with maximum plasma concentrations between 1.2 and 5.4 times higher and areas under the curve between 1.3 and 7.2 times higher, while noting that every one of those studies carried industry funding or industry-employed authors [1,2,3,4]. A powdered liposomal preparation produced without organic solvents has separately been shown to deliver significantly greater absorption than the unencapsulated vitamin in a randomised double-blind crossover design [5].

Manufacturers seeking to support such claims frequently commission preclinical rodent pharmacokinetic studies, which are faster, cheaper and less regulated than human trials. The implicit assumption is that a rat can act as a proxy for a human in detecting differences in oral vitamin C delivery. That assumption deserves more scrutiny than it usually receives.

Humans, other primates, guinea pigs and certain bats cannot synthesise ascorbate, having lost a functional L-gulonolactone oxidase, the terminal enzyme of the biosynthetic pathway. This is precisely why vitamin C is a vitamin for these species and not for others. Rats retain the enzyme and synthesise ascorbate continuously in the liver, maintaining plasma concentrations without any dietary contribution at all. An orally administered dose in a rat is therefore superimposed on a large, actively regulated endogenous pool, whereas in a human the same dose is superimposed on a pool that is entirely diet-derived.

The consequence is straightforward in principle and rarely tested in practice: if the endogenous contribution is large relative to the absorbed dose, the measured plasma signal will be dominated by synthesis rather than by absorption, and the study will be unable to distinguish between formulations however well or badly they deliver. The test of this is an untreated control arm. If untreated animals show plasma exposure comparable to dosed animals, the model has no resolving power and no comparison made within it is interpretable. Such an arm is commonly omitted, because the comparison of interest is between formulations rather than against nothing.

This paper reports a four-arm single-dose oral pharmacokinetic study of a liposomal vitamin C powder in Sprague Dawley rats in which an untreated arm and an unencapsulated active comparator were both included. The liposomal formulation showed a consistent directional advantage over the comparator. The untreated arm showed that the comparator itself was indistinguishable from no treatment. We report both findings, and argue that the second governs the interpretation of the first.

2. Materials and Methods

2.1. Test Article

The test article was a spray-dried liposomal vitamin C powder manufactured by Bionutricia Manufacturing Sdn. Bhd. under Malaysian liposomal encapsulation patent application PI2023005773, supplied as a yellowish powder and stored at 2 to 8 degrees Celsius. The positive comparator was a commercially available over-the-counter ascorbic acid product. The negative control was reverse osmosis water. Both the test article and the comparator were freshly prepared on the day of dosing and reconstituted in reverse osmosis water to the required concentrations, at 100 and 200 milligrams per millilitre for the 1000 and 2000 milligram per kilogram arms respectively.

The test facility recorded that identity, strength, stability and chemical composition of the test article were not independently characterised within the study, and that batch number, production date and expiry date were not supplied. This is a limitation of the dataset and is treated as such throughout.

2.2. Physicochemical Characterisation

Mean vesicle diameter and zeta potential of the liposomal preparation were determined by dynamic light scattering at the School of Pharmacy, Lincoln University College, Selangor, Malaysia. Polydispersity index was not determined and is therefore unavailable.

2.3. Animals and Husbandry

The study was performed at the Industrial Biotechnology Research Centre, Block 19, SIRIM Berhad, Shah Alam, Selangor, Malaysia, under study report R1082/23/B19/204, sponsored by Bionutricia Manufacturing Sdn. Bhd. The facility is compliant with Good Laboratory Practice; the study was conducted at that facility but was not required to comply with Good Laboratory Practice principles, and this is stated in the report.

The study was approved by the SIRIM Institutional Animal Care and Use Committee under approval number SIRIM-IACUC/B19-204/0047, dated 29 September 2023.

Twenty-eight specific pathogen free male Sprague Dawley rats were obtained from Interscience Sdn. Bhd., Shah Alam, Selangor, of which twenty-four were dosed and four held as spares. Animals were 8 to 9 weeks of age and weighed 240 to 350 grams at dosing. They were examined by a veterinarian on arrival, quarantined for five days and acclimatised for six, with veterinary oversight by a visiting veterinary officer throughout.

Rats were housed singly in individually ventilated cages on laboratory and corn cob bedding, with chew stick enrichment, under a 12 hour light and 12 hour dark cycle at 22 plus or minus 3 degrees Celsius and 30 to 70 per cent relative humidity, recorded by data logger. Animals received Altromin rodent maintenance diet as 10 millimetre pellets and reverse osmosis drinking water without restriction. Vital signs were verified within normal limits before any procedure. Animals were fasted overnight before dosing with free access to water, and fed normally afterwards. Dosing took place on 11 October 2023 and all animals were terminated at approximately 24 hours post-dose, with necropsy limited to gross observation of the vital organs. Histopathology was not performed, at the sponsor's request.

2.4. Study Design and Dosing

Animals were allocated to four oral treatment arms as set out in Table 1. Dose volume was 1 millilitre per 100 grams of body weight, calculated from the weight recorded on the day of dosing and delivered by round-headed stainless steel gavage tube.

2.5. Blood Sampling and Bioanalysis

Whole blood was collected from the tail vein at seven time points: pre-dose and 0.5, 1, 2, 4, 8 and 24 hours after dosing, taking 0.2 to 0.25 millilitres per intermediate point and 0.5 millilitres at termination. Samples were drawn into potassium ethylenediaminetetraacetic acid tubes, inverted to mix, held on wet ice or at 2 to 8 degrees Celsius, centrifuged at 2000 times gravity for 10 minutes at room temperature, and the plasma transferred to cryovials and stored at minus 80 degrees Celsius before analysis by liquid chromatography coupled to mass spectrometry. No acidic stabiliser is recorded in the sample handling procedure; the implications are addressed in the Discussion.

2.6. Pharmacokinetic and Statistical Analysis

Non-compartmental analysis was performed by the test facility on the plasma concentration against time profiles. Because elimination phase data were absent across all arms, half-life, clearance and volume of distribution could not be derived, and the reported parameters are limited to maximum observed concentration, time to that maximum, concentration at the last sampled point, area under the concentration-time curve to the last sampled point, the corresponding first-moment curve, and mean residence time.

Group means, standard deviations and between-arm comparisons reported in Section 3 were recomputed by the present authors directly from the individual animal values, using Welch's two-tailed t-test, which does not assume equal variances. Pharmacokinetic parameters are available for three animals per arm and are analysed at that number; body weight was recorded for all six animals per arm and is analysed at six. The study was not designed or powered as a hypothesis test, and these comparisons are reported for transparency about effect size and dispersion rather than as confirmatory statistics.

3. Results

3.1. Physicochemical Characteristics

The vesicle population is submicron but sits above the 200 nanometre ceiling conventionally applied to nanoscale liposomes, and the near-neutral surface charge indicates that electrostatic repulsion contributes little to colloidal stability. Neither observation implies that the product is unstable. In a dried powder the vesicles are immobilised within the carbohydrate continuous phase and are sterically rather than electrostatically stabilised, a mechanism that is robust in the pack and tested only on reconstitution.

Table 2. Vesicle characteristics of the liposomal vitamin C powder by dynamic light scattering, against the conventional thresholds applied to liposomal delivery systems.

Parameter	Value	Conventional threshold
Mean vesicle diameter	392.7 nm	below 200 nm for nanoscale
Zeta potential	−1.4 mV	beyond ±30 mV for electrostatic stability
Polydispersity index	not determined	0.3 or below for monodispersity

Table 2. Vesicle characteristics of the liposomal vitamin C powder by dynamic light scattering, against the conventional thresholds applied to liposomal delivery systems.

Parameter	Value	Conventional threshold
Mean vesicle diameter	392.7 nm	below 200 nm for nanoscale
Zeta potential	−1.4 mV	beyond ±30 mV for electrostatic stability
Polydispersity index	not determined	0.3 or below for monodispersity

3.2. Tolerability and Body Weight

No mortality, moribundity or evident pain occurred in any arm over the 24-hour exposure. No sign of ill health or overt systemic toxicity was observed at one hour or four hours after dosing, and no abnormal behaviour was recorded during cage-side observation. Gross necropsy at termination disclosed no abnormalities in any selected organ in any arm, including at the 2000 milligram per kilogram liposomal dose.

Body weight was recorded for all six animals in each arm and declined in every animal over the 24 hours following administration, as expected after an overnight fast and repeated blood sampling. The magnitude differed between arms (Table 3). Animals given 2000 milligrams per kilogram of the liposomal preparation lost 22.3 grams on average, approximately half as much again as the untreated animals at 14.6 grams, with the largest individual loss in the study at 38.4 grams. The difference did not reach conventional significance (p = 0.09), and neither the 1000 milligram per kilogram liposomal arm nor the ascorbic acid arm differed from untreated animals.

3.3. Pharmacokinetic Parameters

Individual and summary parameters for all four arms are given in Table 4. Dispersion is wide in every arm, and in the two 2000 milligram per kilogram arms the individual areas under the curve span a factor of two and seven respectively.

3.4. The Liposomal Formulation Against the Active Comparator

At the matched 2000 milligram per kilogram dose, the liposomal arm exceeded the ascorbic acid comparator on both exposure measures, by 3.08 times on maximum concentration and 2.14 times on area under the curve. Recomputed against the observed dispersion at three animals per arm, neither difference reached conventional statistical significance, at p equal to 0.19 and 0.18 respectively.

3.5. Both Treated Arms Against the Untreated Control

The untreated control arm is the decisive comparison and is set out in Table 5. Animals given 2000 milligrams per kilogram of commercial ascorbic acid recorded a mean area under the curve of 1.83, against 1.65 in animals given reverse osmosis water alone: a ratio of 1.11, with p equal to 0.90. Animals given 1000 milligrams per kilogram of the liposomal preparation recorded 0.69, which is 0.42 times the untreated value. The liposomal arm at 2000 milligrams per kilogram exceeded the untreated animals by 2.38 times on area under the curve, which again did not reach significance at p equal to 0.16.

3.6. Timing and Absolute Magnitude of the Plasma Signal

Maximum concentration occurred at the terminal 24-hour sampling point in eight of the twelve animals analysed, including all three animals in the ascorbic acid arm and two of three in the untreated arm. Plasma concentrations were therefore still rising when sampling stopped, the reported maxima are not true maxima, and the areas under the curve truncate an unknown fraction of total exposure. No elimination phase was captured in any arm, which is why half-life and clearance could not be derived.

Across every arm, controls included, individual plasma concentrations at all time points lay between zero and 0.63 micrograms per millilitre. Normal rat plasma ascorbate is generally reported in the region of ten to twenty micrograms per millilitre. The measured values are therefore approximately one to two orders of magnitude below the expected physiological baseline.

4. Discussion

4.1. The Untreated Control Invalidates the Comparison It Was Included to Support

A two gram per kilogram oral dose of ascorbic acid is enormous. Scaled allometrically it is far beyond any human supplemental intake. That dose produced plasma exposure in these animals that was statistically indistinguishable from giving them water, at a ratio of 1.11 and a probability of 0.90. A dosing regimen that cannot distinguish two grams per kilogram of the reference substance from nothing at all cannot distinguish between two formulations of it.

The explanation is not obscure. Rats synthesise ascorbate continuously through a functional L-gulonolactone oxidase and regulate their plasma concentration homeostatically. An oral dose enters a pool that is already full and actively defended, and the surplus is cleared renally rather than accumulating. In a human, who has no biosynthetic capacity, the same dose enters a pool whose entire content is diet-derived, and the absorbed fraction is visible against a much quieter background. The rat is not a scaled-down human for this analyte; it is a different system.

This is why the absence of an untreated arm is so damaging in published rodent work on vitamin C formulations. Without it, two treated arms will differ by some amount, a ratio will be computed, and the study will appear to have measured oral delivery. It will in fact have measured the difference between two numbers drawn from a distribution dominated by hepatic synthesis. The untreated arm is what reveals that, and it is the arm most often omitted.

4.2. A Weight Signal at the Top Dose

The one endpoint in this study recorded in all six animals per arm, rather than three, is body weight, and it is the only comparison with even modest power. Animals at the top liposomal dose lost half as much weight again as untreated animals over 24 hours. The comparison falls short of significance and every animal in the study lost weight, so this is a signal rather than a finding. It is nonetheless the kind of signal worth carrying forward: a two gram per kilogram bolus of a lipid and carbohydrate matrix is an appreciable osmotic and gastrointestinal load, and the same lipid envelope that is argued to protect the gastric mucosa from unencapsulated ascorbate also delivers material that must itself be handled. Gross necropsy was unremarkable and no histopathology was performed, so nothing further can be said from these data. Any repetition should record food and water intake alongside weight, and should retain the full complement of animals through to analysis.

4.3. What the Liposomal Result Does and Does Not Establish

The liposomal formulation at the matched dose exceeded the comparator on both exposure measures, and the direction was consistent across maximum concentration, area under the curve and the comparison against untreated animals. We regard this as encouraging and as compatible with enhanced delivery. We do not regard it as established.

Three reasons. First, neither difference reached significance at three animals per arm, with probabilities near 0.2; the study was not powered for this test and the variability between individual animals is large, with areas under the curve differing sevenfold within a single arm. Second, the dose-response relationship is incoherent: the liposomal arm at 1000 milligrams per kilogram fell below the untreated animals, which cannot be a pharmacological effect of the test article and is better read as noise in a system with poor resolving power. Third, and decisively, the comparator arm was itself indistinguishable from no treatment, so the ratio of liposomal to comparator is a ratio against a baseline that carries no information about oral delivery.

A positive result obtained inside a model that fails its own negative control is not a positive result. It is an uncontrolled observation that happens to point in the direction the sponsor hoped for, and it should be reported with that qualification attached rather than without it.

4.4. The Absolute Concentrations Are Not Physiologically Plausible

Values one to two orders of magnitude below the accepted baseline for rat plasma ascorbate require an explanation, and the most parsimonious one is pre-analytical. Ascorbate in plasma oxidises rapidly to dehydroascorbate and then irreversibly to diketogulonate unless the sample is acidified, typically with metaphosphoric acid, or protected with a reducing agent at the point of separation. The recorded procedure describes anticoagulation, chilling, centrifugation and storage at minus 80 degrees Celsius, but no acidic stabiliser. Freezing arrests but does not reverse oxidation that has already occurred during collection and processing.

If the plasma was unstabilised, the measured concentrations represent a variable surviving fraction of the true ascorbate content rather than the content itself, and the between-animal variability observed here is at least partly a variability in sample handling. The consequence for interpretation is specific: ratios between arms processed identically retain some meaning, absolute concentrations do not, and no comparison with values from other laboratories is admissible. We have confined our conclusions accordingly, and we recommend that this detail be established and reported in any repetition.

4.5. Truncated Sampling

Maximum concentration at the final sampling point in two-thirds of animals indicates that the sampling window closed before the curve turned over. For an orally dosed water-soluble vitamin with rapid renal clearance, a genuine maximum at 24 hours is not plausible as an absorption phenomenon; it is more consistent with a slow drift in the endogenous pool, or with a diurnal component, than with the administered dose. Either way the reported maxima are lower bounds, the areas under the curve are truncated, and no elimination parameter can be derived. Future work should extend sampling until a decline is observed, or should abandon the terminal point as uninformative.

4.6. What Should Be Done Instead

For a formulation whose entire value proposition is oral bioavailability, the test system must be capable of detecting oral bioavailability. Three options are available and all are preferable to repeating this design in rats.

• — A species that cannot synthesise ascorbate. The guinea pig is the conventional choice, has a long history in vitamin C research for exactly this reason, and would allow a dose-response relationship to be established against a genuinely quiet baseline.
• — A rat strain lacking the biosynthetic pathway. The osteogenic disorder Shionogi strain carries a defective L-gulonolactone oxidase and is scorbutic without dietary supplementation, retaining rodent convenience while removing the confound identified here.
• — A human crossover trial. This is the only design that answers the commercial question directly, is the design used in the published studies against which any new formulation will be compared, and avoids the interspecies extrapolation entirely. For a formulation already in commerce and generally recognised as safe, the regulatory and ethical burden is modest.

Whichever is chosen, an untreated arm should be retained. Its function is not to provide a comparison of interest but to demonstrate that the model has the resolving power to make one.

5. Limitations

Six animals were dosed in each arm and are individually identified in the source report, but pharmacokinetic parameters are reported for only three per arm; the plasma results for the remaining nine animals do not appear in the report and all pharmacokinetic analyses here are therefore at three. Body weight, recorded for all six, is analysed at six. The study was neither designed nor powered as a hypothesis test and the comparisons reported here are descriptive. No elimination phase was captured, so half-life, clearance and volume of distribution are unavailable and the areas under the curve are truncated. The test article was not independently characterised for identity, strength or composition within the study, and batch identification was not supplied. Plasma handling is not documented as including an acid stabiliser, which places the absolute concentrations in question as discussed. Dynamic light scattering was performed on the formulation but no polydispersity index was obtained, and vesicle integrity after reconstitution and after gastric transit was not assessed. Only male animals were used. The proposed role of endogenous synthesis is inferred from the untreated control and from established species physiology rather than demonstrated directly; measurement of hepatic L-gulonolactone oxidase activity, or repetition in a non-synthesising species, would test it.

Several internal inconsistencies in the source report should be recorded. The animals are identified as male in the test system description, the cage allocation table and the dosing record, but as female in the four body weight tables; male is taken as correct, being stated in three independent places including the strain description. The cage allocation table labels the third arm as 1000 milligrams per kilogram, whereas the study design table, the dosing record, the body weight table and the pharmacokinetic appendix all give 2000; the latter is taken as correct. The negative control vehicle is given as reverse osmosis water in the study design and as saline in the dosing record. The dosing record carries an administration date one day later than the study timetable, which would place dosing and necropsy on the same day and is inconsistent with the 24-hour design. The negative control body weight table carries a dose concentration and test article mass copied from an active arm, although that arm received vehicle alone. None of these affects the plasma or body weight values analysed here, but each is noted for completeness.

6. Conclusions

In a four-arm single-dose oral study, a liposomal vitamin C powder produced approximately three times the maximum plasma concentration and twice the area under the curve of an equal dose of commercial ascorbic acid in Sprague Dawley rats. Neither difference reached statistical significance at three animals per arm, and the dose-response relationship was not monotonic.

The untreated control arm governs the interpretation. Two grams per kilogram of unencapsulated ascorbic acid produced plasma exposure indistinguishable from reverse osmosis water. A model in which the reference substance is undetectable at an extreme dose cannot measure the bioavailability of anything, and the apparent advantage of the liposomal formulation within that model is therefore uninterpretable as evidence of enhanced delivery, whatever its direction.

We draw two conclusions for the field. Rodent pharmacokinetic studies of vitamin C formulations conducted in ascorbate-synthesising species should be regarded as uninformative unless an untreated arm demonstrates that the model can resolve the dose, and such an arm should be mandatory rather than optional. And manufacturers seeking to substantiate bioavailability claims should go directly to a non-synthesising species or to human crossover, since the rodent study that appears to be the cheap first step may answer no question at all.