Comparative evaluation of USP II, USP IV, TIM-2, and the dynamic colon model as surrogates for human colonic drug dissolution

Abstract

This study compares conventional dissolution apparatuses (USP II and IV) with advanced biorelevant models (TIM-2 and the Dynamic Colon Model, DCM) for assessing colonic drug dissolution. Theophylline dissolution from modified-release Uniphyllin Continus® tablets was evaluated across varying hydrodynamic conditions, with the DCM serving as a physiologically validated benchmark. Dissolution was highly dependent on USP II paddle speed (50–200 rpm). In the USP IV increasing flow rate (4–50 mL min⁻¹) increased 24 h dissolution but had relatively low impact at 1–5 h. In both systems, the lowest agitation conditions (50 rpm and 4 mL min⁻¹) provided the closest agreement to DCM profiles. A simplified TIM-2 configuration was used to isolate hydrodynamic effects, demonstrating motility-dependent release, with slower mixing producing dissolution profiles most similar to the DCM. Incorporation of dewatering improved agreement with the stimulated DCM condition for all systems. The consistent trend across systems of improving similarity to the DCM data with decreasing agitation suggests that conditions below the ranges tested here may be warranted in future studies. USP II and USP IV remain practical and robust tools for quality control but TIM-2 and the DCM better replicate colonic hydrodynamics. TIM-2, even in its simplified configuration in this study, offers additional capability to modulate peristaltic motility patterns and provides a promising platform into which the standard TIM colonic fluids, secretion and absorption functions, can be progressively reintroduced to improve biopredictivity. These findings highlight the complementary role of conventional and advanced dissolution approaches and support integrating physiologically relevant parameters into dissolution testing frameworks to improve biopredictive assessment for modified release formulations.

Lower hydrodynamic conditions in biorelevant dissolution models more closely reproduced colonic drug release than conventional USP methods.

1. Introduction

Colonic drug delivery systems (CDDS) are specialised formulations designed to target the colon as the site of drug release. The most commonly used technologies are hydrophilic matrix systems due to their ability to consistently reproduce desirable drug release profiles while offering cost-effective manufacturing solutions.¹

The combined mechanisms of diffusion, polymer swelling and erosion control drug release from matrix-based controlled release systems.² The colon exhibits distinct hydrodynamic conditions, such as limited fluid availability and reduced mechanical agitation, which can hinder these processes.³

Simulating colonic conditions presents several challenges due to the unique physiological environment of the colon. Unlike other regions of the gastrointestinal tract, the colon has a high capacity for water absorption, and its contents are highly viscous and poorly mixed, which restricts drug diffusion and access to the absorptive membrane.⁴ The impact of food on dissolution in simulated gastric fluid has demonstrated that the increase in viscosity reduces the rate of drug release.⁵ In a digital model of the in vitro dynamic colon model (DCM) the impact of increased viscosity also showed a reduced rate of drug dissolution.⁶

To study drug release under controlled conditions, several dissolution models have been developed, ranging from standard compendial apparatuses to advanced biorelevant systems. While these models provide valuable insights into formulation performance, they often cannot fully mimic the dynamic physiological conditions of the colon, such as dewatering, which can limit polymer swelling, gel-layer formation, and drug solubilisation, thereby altering drug release behaviour. Conventional dissolution tests typically use constant volumes and viscosities, potentially overestimating release from modified-release formulations that depend on water availability for swelling and erosion. Although dewatering has not yet been incorporated into published dissolution methods, it is likely to be an important factor for accurately reproducing colonic conditions and predicting in vivo performance.⁷

Several dissolution equipment types are defined by the United States Pharmacopeia (USP), each appropriate for various formulations and testing requirements.¹ For colonic testing the most commonly used apparatus are USP types I, II, III and IV. For each system the hydrodynamics are controlled by the stirring, dipping or flow rate. USP IV has previously been reported to have more in vivo-relevant hydrodynamics compared to other USP apparatus.⁸ In USP Apparatus IV (flow-through cell), hydrodynamics are governed primarily by the flow rate of the dissolution medium, as well as by cell geometry and flow conditions. The system may be operated with or without the inclusion of glass beads within the flow cell. Glass beads are commonly introduced to act as a flow distributor, promoting more uniform fluid movement around the dosage form. In contrast, operation without beads typically results in less uniform flow patterns, which may enhance localised shear and contribute to increased matrix erosion and drug release rates.⁹ Although glass beads are often described as promoting laminar flow, experimental studies have demonstrated that the flow field is typically heterogeneous and not fully developed laminar, with recirculation and local velocity variations observed under standard operating conditions.¹⁰

Designing physiologically relevant in vitro tests for modified-release formulations requires replication of the hydrodynamic conditions governing diffusion, swelling, and erosion. In matrix tablets, drug release is largely controlled by the thickness of the gel layer formed between the diffusion and erosion fronts, which is influenced by fluid flow and mechanical shear. Previous work has shown that the Dynamic Colon Model (DCM) mimics the human colon¹¹ and further that the data generated in this model was superior to USP II dissolution (at 25, 50 and 100 rpm) in reproducing in vivo release profiles.¹² The DCM offers a robust reference for assessing formulation performance and enhancing in vitro–in vivo correlation (IVIVC).¹³ Previous studies using the Dynamic Colon Model (DCM) applied MRI-derived colonic motility profiles from healthy volunteers and stimulated conditions induced by PEG or maltose ingestion.^14–16 For theophylline matrix tablets, stimulated profiles produced faster and more complete drug release (>90% over 24 h) than healthy baseline conditions (57% over 24 h). Power law modelling showed an increase in the release exponent from 0.74 to 0.90 under stimulated conditions, indicating a shift towards erosion-controlled release due to higher hydrodynamic shear. Coupling DCM dissolution data with physiologically based pharmacokinetic modelling demonstrated that healthy motility profiles accurately predicted in vivo performance for a 200 mg dose, whereas maltose-stimulated conditions overpredicted systemic exposure.¹² At a higher dose (400 mg), however, more aggressive erosion was required to reproduce observed clinical data.¹² As the DCM is not currently commercially available, there remains a need for alternative dynamic dissolution systems that can support wider adoption of advanced, biorelevant testing approaches and strengthen in vitro–in vivo correlations.

The TIM-2 system is a dynamic, in vitro model of the human large intestine, usually inoculated with a dense and metabolically active colon microbiota of human origin and maintained as TIM Colonic Fluid (TCF) through continuous delivery of pre-digested substrate from the ileum, colonic secretions and filtration. It replicates key physiological conditions, including body temperature, dynamic luminal pH and strict anaerobiosis.¹⁷ The system also simulates mixing and transit of intestinal contents, water absorption, and the removal of metabolites and other low molecular weight compounds via a semipermeable membrane. This enables controlled and reproducible studies of colonic processes under biorelevant conditions.¹⁸ The TIM-2 model consists of four interconnected glass compartments, each containing a flexible membrane. Temperature-controlled water is circulated between the rigid glass wall and the flexible membrane, allowing controlled deformation of the system. By applying pressure to this water at regular intervals and in a defined sequence, the membranes expand and contract to generate peristaltic waves, which mix the luminal contents and propel them through the system.¹⁷ Although the TIM-2 apparatus has been used to assess dissolution of drugs,¹⁹ there has yet to be a study that explores the use of TIM-2 as a closed dissolution apparatus to exploit the biorelevant hydrodynamics generated in this apparatus.

This study aims to compare the dissolution profiles of theophylline from a controlled-release matrix tablet, Uniphyllin Continus 200 mg, across four different dissolution systems: USP Apparatus II, USP Apparatus IV, TIM-2, and the Dynamic Colon Model (DCM). The impact of dewatering within these systems was evaluated with reference to physiological dewatering data. The comparison evaluated differences in drug release behaviour over a 24 h period, with theophylline data from the DCM serving as a physiologically relevant benchmark.³ The study seeks to further assess the predictive value of these in vitro systems for human colonic performance.

2. Materials and methods

2.1. Products and chemicals

Uniphyllin Continus 200 mg tablets, containing theophylline as the active pharmaceutical ingredient, were obtained from a local pharmacy and represent a modified-release dosage form; all other chemicals (potassium sodium dibasic; sodium phosphate monobasic monohydrate; 0.1 M hydrochloric acid; sodium carboxymethylcellulose (NaCMC); glacial acetic acid; acetonitrile) were purchased from Sigma Aldrich and were of analytical or HPLC grade.

2.2. Dissolution conditions and analysis

The dissolution conditions were designed to match existing in vitro data from the DCM that showed motility patterns derived from stimulated conditions in vivo increased theophylline release (>90% vs. 57% at 24 h) and shifted the mechanism towards erosion-driven release due to higher shear;¹² this resulted in adaptations to the TIM-2 apparatus that is further discussed in section 2.3.3.

2.2.1. Dissolution media composition

Phosphate buffer at pH 7.4 was used as the dissolution medium, prepared by dissolving 20.214 g of potassium sodium dibasic and 3.397 g of sodium phosphate monobasic into 800 ml of distilled water. The pH was adjusted to 7.4 by addition of 0.1 M HCl then made up to 1 L with addition of distilled water. The dissolution media volumes were 1000 mL (USP II), 200 mL (USP IV and TIM-2). The dissolution medium in the DCM in the comparative study was 100 mL phosphate buffer solution with 0.25% (w/w) sodium carboxymethylcellulose (NaCMC).³

The dewatering experiments used increasing concentrations of NaCMC to alter the viscosity of the dissolution medium. Concentrations were increased stepwise from buffer (with 0% NaCMC) at the start, then at 1 h (0.25% w/v NaCMC), 2 h (0.50% w/v NaCMC), and 3 h (0.75% w/v NaCMC). Increased concentrations were achieved by removing a fraction of the existing medium at set intervals and replacing it with a more concentrated NaCMC medium to generate the media profile stated above (1 h (0.25% w/v NaCMC), 2 h (0.50% w/v NaCMC), and 3 h (0.75% w/v NaCMC), see Table 1).

Table 1. Stepwise modification of dissolution medium composition during dewatering experiments using the USP II apparatus.

Time (h)	Volume withdrawn (mL)	NaCMC concentration of replacement solution (% w/v)	Resulting bulk NaCMC concentration (% w/v)
0	—	—	0.00
1	250	1.00	0.25
2	250	1.25	0.50
3	250	1.50	0.75

2.2.2. Sampling timepoints

Samples were withdrawn at 1, 2, 3, 4, 5, and 24 hours for all USP and TIM-2 experiments. The theophylline concentration within samples was determined using HPLC analysis as described in section 2.2.3.

2.2.3. Analytical methods

Theophylline concentrations were determined using a Thermo Finnigan HPLC system equipped with an X-Select C18 column (5 µm, 150 × 4.6 mm i.d.) maintained at 25 °C. The mobile phase consisted of (A) water adjusted to pH 4.2 with glacial acetic acid (0.05% v/v) and (B) acetonitrile, mixed in a 50 : 50 v/v ratio. The flow rate was 1.0 mL min⁻¹, and detection was carried out at 272 nm. The injection volume was 20 µL for each data point and the total run time was 10 minutes for each sample.

2.3. Dissolution apparatuses

2.3.1. Dissolution using the USP II (paddle) apparatus

Dissolution tests were performed using a USP II (Copley Scientific, Nottingham, UK) paddle apparatus. The temperature of each vessel was maintained at 37 °C. Paddle speeds of 50, 100, 150 and 200 rpm were used for four separate experiments. Sink conditions were maintained throughout the dissolution study to ensure that the drug concentration in the dissolution medium remained well below its saturation solubility (solubility was 6.92 mg mL⁻¹ at pH 7.4),²⁰ thereby allowing for accurate and consistent measurement of drug release. While high paddle speeds are not reflective of physiological conditions, their inclusion allowed assessment of dissolution behaviour under extreme hydrodynamic stress and ensured continuity with previously reported datasets.^3,13

For the dewatering experiments, the dissolution media was modified in a stepwise manner to simulate progressively increasing intestinal viscosity. The experiment was initiated with 1000 mL of dissolution medium containing no NaCMC. At hourly intervals between 0 and 3 h, 250 mL of medium was removed and replaced with a more concentrated NaCMC solution, such that the bulk NaCMC concentration was 0.25% w/v NaCMC at 1 h, 0.50% w/v NaCMC at 2 h, and 0.75% w/v NaCMC at 3 h. The replacement volumes and their corresponding NaCMC concentrations are summarised in Table 1. No further medium changes were made thereafter, and the dissolution was continued under these conditions until a total duration of 24 h. It was assumed that complete mixing was achieved between each addition.

2.3.2. Dissolution using the USP IV (flow through cell) apparatus

The USP Apparatus IV dissolution tests were conducted using a flow-through cell system (CE 7smart, SOTAX AG, Allschwil, Switzerland), which features 22.6 mm-diameter cells. A water bath temperature of 37 °C was maintained throughout the procedure. The pH 7.4 phosphate buffer served as the dissolution medium and was circulated (200 mL volume) in a closed-loop system. The experiment was conducted using 1 g of glass beads within the inner chamber to promote more uniform flow. Theophylline Uniphyllin tablets of 200 mg were then placed into tablet holders located above the glass beads for testing, as shown in Fig. 1. Four different flow rates were used for drug release testing: 4, 16, 25 and 50 mL min⁻¹. Although lower flow rates are generally considered more physiologically relevant for modified-release formulations, higher flow rates were incorporated to examine the sensitivity of drug release to enhanced fluid renewal and to characterise dissolution behaviour across an extended hydrodynamic range.

Fig. 1. Image shows a cartoon of a flow through cell with glass beads.

For the USP IV dewatering experiments, the dissolution media was changed as follows: initially, the dissolution medium consisted of 200 mL of phosphate buffer (pH 7.4); concentrations were increased stepwise from buffer (with 0% NaCMC) at the start, then at 1 h (0.25% w/v NaCMC), 2 h (0.50% w/v NaCMC), and 3 h (0.75% w/v NaCMC). The concentration of NaCMC was increased by removal of 50 mL of the medium at each hour, which was replaced with a more concentrated solution of NaCMC such that the overall concentration was maintained. For example, at 1 h, 50 mL of medium was removed and replaced with 50 mL of 1.0% NaCMC to produce a total concentration of 0.25% NaCMC; at 2 h, 50 mL of medium was removed and replaced with 50 mL of 1.25% NaCMC; and at 3 h, 50 mL of medium was removed and replaced with 50 mL of 1.50% NaCMC. It was assumed that complete mixing was achieved between each addition. No further adjustments were made thereafter, and the experiment continued under these conditions for a total duration of 24 h.

2.3.3. Dissolution using the TIM-2 apparatus

Dissolution studies were conducted using the TIM-2 (InnoGI Technologies, Delft, The Netherlands), shown in Fig. 2. This experimental protocol differs from the standard TIM-2 protocol described by Venema (2015).¹⁷ A pH 7.4 phosphate buffer (as used in the USP apparatus) maintained at 37 °C was used in place of the TCF. Consequently, dynamic pH control and the filtration unit (that usually sits across units 2 and 3 in Fig. 2b) normally required to remove acidic microbial metabolites were not required, nor were the anaerobic conditions normally used to maintain the microbiota.

Fig. 2. Shows (a) TIM-2 apparatus and (b) schematic of the TIM-2 apparatus. Compartment key: 1 = back left compartment; 2 = front left compartment; 3 = front right compartment; and 4 = back right compartment.

While these adaptations simplified the system, they may also have affected the hydrodynamic conditions, particularly the removal of the filter, which typically provides an absorptive sink via a hollow fibre membrane to mimic colonic absorption.

Under these operating conditions, the volume of fluid used was 200 mL, which provided practical operating conditions and matched the volume used in the USP IV apparatus. A sinker was used to secure the tablet, preventing it from freely floating or moving throughout the medium. This ensured consistent positioning of the dosage form, promoting uniform and reproducible drug release during the dissolution test.

Bespoke motility patterns were generated for the TIM-2 apparatus to reflect slow, moderate and fast mixing conditions. The slow, moderate and fast protocols applied approximately 5, 12 and 30 peristaltic contractions per minute, respectively.

For the dewatering experiments, the dissolution media were changed as follows: NaCMC concentrations were increased stepwise from buffer (with 0% NaCMC) at the start, then at 1 h (0.25% w/v NaCMC), 2 h (0.50% w/v NaCMC), and 3 h (0.75% w/v NaCMC). The concentration increase of NaCMC was achieved by the removal of 50 mL of the medium at each hour, which was replaced with the same volume of a more concentrated NaCMC solution. Explicitly, at 1 h, 50 mL of medium was removed and replaced with 50 mL of 1.0% NaCMC to produce a total bulk concentration of 0.25% NaCMC; at 2 h, 50 mL of medium was removed and replaced with 50 mL of 1.25% NaCMC; and at 3 h, 50 mL of medium was removed and replaced with 50 mL of 1.50% NaCMC. It was assumed that complete mixing was achieved between each addition. No further adjustments were made thereafter, and the experiment continued under these conditions for a total duration of 24 h. The concentration of drug removed within this process was quantified and accounted for within the analysis. The site where drug was removed was distant from the location of the tablet to minimise disruption to the hydrodynamics of the system, and it was assumed that mixing of the fresh viscous medium with the existing medium produced a homogeneous dissolution medium prior to the next sample point.

2.3.4. Dissolution using the dynamic colon model

The Dynamic Colon Model (DCM) dissolution data were extracted from published research.³ For further information on the DCM, including images of the apparatus, the reader is directed to preceding papers.^3,11–13 Previous work converted in vivo motility patterns extracted from published MRI studies into motility patterns in the DCM representing the human adult colon under static, baseline, antegrade cyclic propagating pressure wave (CPPW), retrograde CPPW, PEG-stimulated and maltose-stimulated motility, as described in work by O'Farrell et al. 2022.³ The cumulative dissolution profile of a 200 mg Uniphyllin Continus® prolonged-release theophylline tablet over the course of 24 h was measured in the DCM under each motility pattern.³ Further work then integrated these in vitro data from the DCM into physiologically based biopharmaceutics modelling tools to demonstrate the predictive power of the DCM and its superiority to USP II in predicting in vivo exposure of theophylline.¹² For the purposes of this study, the cumulative dissolution profiles generated in the DCM were used as a benchmark, with both the most rapid profile (maltose-stimulated) and the baseline used as comparators to the dissolution data generated and presented in this study using USP II, USP IV and TIM-2.

2.4. Dissolution replicates and statistical treatment

Dissolution data from the USP II, USP IV and TIM-2 experiments were compared to the DCM data under the baseline and stimulated profiles using the model-independent f₁ (difference factor) and f₂ (similarity factor) statistics.²¹ While regulatory guidance typically recommends the use of 12 units per formulation, this study employed four replicates, and the validity of this approach was supported by low variability across replicates, as assessed by the coefficient of variation (CoV). For each time point, the mean and standard deviation were calculated across the four replicates. CoV was used to assess data reliability, with thresholds set at ≤20% for early time points and ≤10% for later time points, in accordance with EMA guidance.

2.5. Model fitting of dissolution data

Equations that describe the release profile as a function of certain characteristics associated with pharmaceutical dosage forms make it simpler to quantitatively evaluate the findings obtained in dissolution studies.¹⁸

Historically, the rate of drug release from polymeric matrices has been calculated using the power law model by Korsmeyer and Peppas (1983)¹⁹ (eqn (1)).

1

Where, the ratio of M_t to M_∞ denotes the fraction of drug remaining within the dosage form (i.e., undissolved drug) at time t, and K represents a kinetic constant that incorporates contributions from both the micro- and macro-structural properties of the system, including polymer composition, matrix porosity, and geometry of the dosage form.¹⁹ This model was chosen because it accurately depicts the non-linear behaviour of drug release from hydrophilic matrix systems, where release kinetics are influenced by diffusion, erosion, and swelling.

Power models were applied to experimental data in Microsoft Excel and Python (v3.13), to enable comparison of dissolution profiles, and characterisation of the release kinetics. Trend lines were created using non-linear regression analysis according to a power-law equation of the type y = ax^b, where x is time, a and b are empirically determined constants, and y is the cumulative percentage of medication released (the ratio of M_t to M_∞). As per best practice these models were applied to fit the early data points where appropriate and the full data set was used where the correlation coefficient was >0.9.

3. Results

3.1. Calibration data

The calibration curves for quantification of theophylline were linear over the concentration range 0.1–0.01 mg mL⁻¹, with a correlation coefficient (R²) of 1.00.

The limit of detection (LOD) and limit of quantitation (LOQ) were calculated to be 0.00041 mg mL⁻¹ and 0.00125 mg mL⁻¹ respectively. Since the calculated LOD and LOQ are well below the lowest standard concentration, the method is sufficiently sensitive and suitable for accurate quantification within the studied range.

3.2. Dissolution profiles under standard conditions

3.2.1. Dissolution USP II paddle apparatus

The impact of agitation intensity on theophylline release using the USP II paddle apparatus is shown in Fig. 3. Drug release increased with paddle speed and after 24 h, complete drug release was achieved in all cases, regardless of paddle speed. Data represent the mean of four tablets (n = 4), with error bars showing standard deviation. Fitting of the power function to the data points up to 5 h showed that the rate of release increased with agitation rate and that the release exponent increased indicative of erosion being dominant in the mechanism of release (see Table S1 in SI).

Fig. 3. The dissolution profile of theophylline (% label claim) from modified-release tablets at 50, 100, 150 and 200 rpm in USP II apparatus. Data points show the mean of 4 repeats with the error bars showing the standard deviation. The equations show the power law model fit over the first 5 h.

The dissolution profiles at all paddle speeds showed small standard deviations and coefficients of variation <7% under all conditions, suggesting consistent release behaviour across replicates (Fig. 3). The sensitivity of the release rate to hydrodynamic conditions was reflected in the slightly higher variability at earlier timepoints, while the fastest speed (200 rpm) was the least variable condition overall (mean CoV 1.0%), as release approached completion by 4–5 h. Later stages (>20 h) showed complete release being observed by 24 h regardless of paddle speed.

3.2.2. Dissolution USP IV flow through cell apparatus

Fig. 4 illustrates the effects of flow rate on the release of theophylline in the USP IV. Cumulative theophylline dissolution at 24 h increased with flow rate from 4 to 50 mL min⁻¹. A step change was observed between 16 and 25 mL min⁻¹ rather than a linear dependence; at flow rates of 4 and 16 mL min⁻¹ the tablets did not reach full drug release within the 24 h experiment. Early dissolution (1–5 h) varied comparatively little across the tested flow rates, spanning only 5.2–7.3% at 1 h and 22.5–26.4% at 5 h.

Fig. 4. The dissolution profile of theophylline (% label claim) from modified-release tablets at 4, 16, 25 and 50 mL min⁻¹ in USP IV with glass beads. Data points show the mean of 4 repeats with the error bars showing the standard deviation. The equations show the power law model fit over the first 5 h.

The release exponent was similar for all flow rates and was indicative of erosion controlled release within this apparatus (see Table S2 in SI).

The coefficient of variation exceeded the 20% threshold at 1 h across flow rates of 4–25 mL per minute, and at 2 h for the 25 mL min⁻¹ condition, while values at other conditions and timepoints remained below this threshold.

3.2.3. Dissolution using TIM-2 apparatus

The results shown in Fig. 5, demonstrate that theophylline release increased under the fast mixing protocol and decreased under the slow mixing protocol.

Fig. 5. The dissolution profile of theophylline (% label claim) in TIM-2 using the fast, moderate and slow mixing protocols. Data points show the mean of 4 repeats, with the error bars showing the standard deviation. The equations show the power law model fit over the first 5 hours.

The release exponent was comparable across all mixing protocols, indicative of erosion controlled release within TIM-2 under moderate and fast mixing protocols (see Table S3, SI).

Variability was most pronounced during the early phase (1–5 h), particularly under moderate and slow mixing protocols. The coefficient of variation exceeded 20% for both the slow and fast mixing protocols at the 1 h timepoint. However, variability decreased over time as dissolution progressed and profiles converged towards near-complete release.

3.3. Correlation of dissolution profiles with DCM data

3.3.1. Statistical comparison

The difference factor (f₁) and similarity factor (f₂) were computed to evaluate the comparability of drug release profiles to those measured in the DCM under baseline (Table 2) and maltose-stimulated (Table 3) conditions (f₁ ≤ 15 and f₂ ≥ 50 indicate similarity). None of the USP II profiles met both criteria for similarity to the DCM baseline conditions, however the lowest paddle speed showed the highest similarity, with the f₂ criterion met. The USP IV apparatus met the criteria for similarity to the DCM maltose-stimulated data at 4 and 16 mL min⁻¹, whereas the TIM-2 data did not meet the criteria under any tested condition.

Comparison of drug release profiles from USP II; USP IV and TIM-2 apparatus under varying hydrodynamic conditions using the difference factor (f₁) and similarity factor (f₂), with the DCM baseline data serving as the reference method.

Paddle apparatus (USP II)
Speed (rpm)	Difference factor f1	Similarity factor f2
50	176.7	28.9
100	252.0	22.7
150	341.5	18.0
200	456.2	12.4

Flow through cell (USP IV)
Speed (mL per minute)	Difference factor f1	Similarity factor f2
4	100.1	37.0
16	113.2	34.8
25	141.1	29.9
50	153.3	27.7

TIM-2
Mixing protocol	Difference factor f1	Similarity factor f2
Fast	383.7	14.4
Moderate	260.3	22.5
Slow	170.5	29.8

Comparison of drug release profiles under varying conditions using the difference factor (f₁) and similarity factor (f₂), with the DCM maltose-stimulated data serving as the reference method.

Paddle apparatus (USP II)
Speed (rpm)	Difference factor f1	Similarity factor f2
50	39.5	51.2
100	77.4	36.2
150	122.6	25.5
200	180.4	17.0

Flow through cell (USP IV)
Speed (mL per minute)	Difference factor f1	Similarity factor f2
4	5.0	87.9
16	7.5	84.0
25	21.9	56.6
50	27.7	52.1

TIM-2
Mixing protocol	Difference factor f1	Similarity factor f2
Fast	146.2	20.6
Moderate	83.4	32.6
Slow	37.7	51.8

Comparison of the TIM-2 profiles to USP IV (Table 4) showed that although the f₁ values did not meet similarity criteria, f₂ values exceeded 50 for the USP IV at all flow rates, indicating comparable profile shapes. f₂ values were higher at 16, 25 and 50 mL min⁻¹ compared to 4 mL min⁻¹, suggesting that the effective hydrodynamic conditions in TIM-2 are more comparable to those under flow rates greater than 4 mL min⁻¹. However, this relationship is indirect, and further work is required to better define the correspondence between mixing intensity in TIM-2 and flow dynamics in USP IV and how this may translate to shear on the tablet surface.

Table 4. Comparison of USP IV and TIM-2 profiles using the f₁ and f₂ statistic.

	TIM-2 slow mixing protocol
	Difference factor f1	Similarity factor f2
USP IV (4 mL min−1)	26.7	50.7
USP IV (16 mL min−1)	22.0	54.5
USP IV (25 mL min−1)	20.6	55.7
USP IV (50 mL min−1)	20.2	55.3

3.4. Dewatering experiments

3.4.1. Dissolution data with dewatering implemented

Experiments were conducted to explore the impact of dewatering on theophylline dissolution using USP II (50 rpm), USP IV (4 and 25 mL min⁻¹) and the simplified TIM-2 set-up under moderate and slow motility protocols. Results are shown in Fig. 6.

Fig. 6. Comparison of dissolution profiles of theophylline (% label claim) under dewatering conditions from the USP II; USP IV and TIM-2 apparatus in comparison to the DCM data under baseline and maltose simulated conditions.

Following implementation of dewatering, USP IV and the simplified TIM-2 produced comparable theophylline dissolution profiles to the maltose-stimulated DCM over the first 5 h. In all set ups in this study, dissolution rate and extent did not reach the low levels reported under baseline conditions in the DCM. Comparative f₁ and f₂ values are reported in Table 5.

Table 5. Comparison of drug release profiles using a range of apparatus under dewatering conditions using the difference factor (f₁) and similarity factor (f₂), with the DCM data serving as the reference method.

	Baseline DCM		Maltose stimulated DCM
	Difference factor f1	Similarity factor f2	Difference factor f1	Similarity factor f2
USP II (50 rpm)	117	40.7	69.3	52.0
USP IV (25 mL min−1)	67.1	50.5	12.5	83.0
USP IV (4 mL min−1)	43.2	58.5	7.8	91.2
TIM-2 (moderate)	78.5	48.8	31.6	65.0
TIM-2 (slow)	77.2	49.6	33.3	66.7

None of the dissolution data under simulated dewatering conditions met both criteria for similarity to the DCM baseline data. However, for USP II at 50 rpm, dewatering improved similarity to the maltose-stimulated DCM data, with the f₂ criterion met. This indicates comparable profiles, although the elevated f₁ value indicates large absolute differences at particular time points. In USP IV, dewatering reduced the rate of dissolution, improving similarity to the maltose-stimulated DCM data at 25 mL min⁻¹ (f₁: 21.9 → 12.5; f₂: 56.6 → 83.0), but decreasing similarity at 4 mL min⁻¹ (f₁: 5.0 → 7.8; f₂: 87.9 → 91.2). In TIM-2, dewatering improved the f₁ criterion but reduced the f₂ criterion for both mixing protocols.

4. Discussion

The sensitivity of Uniphyllin Continus theophylline modified-release tablets to hydrodynamic conditions across dissolution systems differing in geometry, fluid volume, and agitation mechanism is consistent with an erosion-controlled release mechanism, as supported by the Korsmeyer–Peppas analysis.

Drug release in USP II was strongly dependent on paddle speed, consistent with previous reports for erodible matrix formulations,^21–23 including studies of the same drug product by O'Farrell et al.³ In the USP IV, flow rates up to 50 mL min⁻¹ were investigated, extending the hydrodynamic range typically investigated in USP IV studies.⁸ The data point to an important mechanistic transition during dissolution in the USP IV. Differences between flow conditions were relatively small during the initial phase (1–5 h) but became increasingly pronounced thereafter. This suggests that release is initially governed by polymer hydration and swelling, with erosion becoming progressively more important once a hydrated gel layer has formed. This interpretation is consistent with observations by Garbacz et al.,²⁰ who showed that HPMC matrix tablets were relatively resistant to mechanical stress during early hydration but became increasingly susceptible to stress-induced release after approximately 5 h, as matrix softening progressed. The absence of a similar effect in USP II may indicate that this transition is more readily observed in systems with lower local fluid volumes surrounding the dosage form. This time-dependent behaviour underscores the importance of evaluating dissolution profiles across the full timecourse (relative to in vivo transit times). As such, early timepoint comparisons alone or accelerated dissolution tests may be misleading.

Neither the DCM nor the in vitro apparatus in this study incorporated an absorptive sink (though TIM-2 does when operated under standard conditions). In vivo, colonic absorption of theophylline continuously maintains a drug concentration gradient that promotes ongoing release from the matrix. In contrast, accumulation of dissolved drug in closed in vitro systems may reduce the driving force for dissolution if sink conditions are not maintained. Similar limitations have been reported for higher-dose theophylline formulations¹² and for quetiapine extended-release tablets,²⁴ where incomplete dissolution was attributed to progressive loss of sink conditions. These considerations may become increasingly important when evaluating higher-dose formulations under non-sink conditions.

Benchmarking against the DCM, where the motility patterns were informed by human colonic motility data, provided insight into the physiological relevance of the investigated dissolution systems. The DCM baseline condition was characterised by relatively low-amplitude contractions of isolated haustra, limited contractile waves propagating multiple adjacent haustra, and periods of inactivity of the mimic colonic walls. In contrast, the maltose-stimulated condition incorporated more frequent isolated and propagating contractions which propagated both forwards and backwards, greater lumen occlusion, faster occlusion rates and also periods of complete inactivity. These differences in motility produced substantially different dissolution behaviour, highlighting the importance of hydrodynamic conditions in governing drug release from modified-release matrix systems.

A practical difference between systems was dosage form mobility. Under baseline DCM conditions, the tablet was reported to remain within a single haustrum throughout the experiment, whereas under stimulated conditions, tablet orientation changed, exposing additional surfaces to fluid shear.³ This highlights the importance of a physiologically relevant agitation mode, not just in creating relevant fluid motion but also in moving the tablet, and should be considered in future experiment design. In this TIM-2 experiment, despite peristaltic wall movement that may be able to reorient the tablet, as in the DCM, the tablet position was fixed using a sinker. The sinker design enabled access to all surfaces of the dissolution medium throughout the experiment. While the effect of tablet movement within the DCM was not quantified directly, it represents an additional physiological variable that is generally absent from conventional dissolution testing and may warrant further investigation in future studies. Despite this, the higher dissolution achieved in TIM-2 compared to the DCM suggests that the impact of dose mobility may be low under higher-intensity mixing conditions.

A further consideration is the homogeneity of the dissolution medium. The DCM study showed that baseline motility was ineffective at homogenising the lumen contents, requiring sampling at multiple locations to characterise the distribution of dissolved drug.¹³ This suggests that under low-motility conditions, particularly when more viscous media are used or viscosity is progressively increased by dewatering, sampling from a single location may be insufficient. Previous studies have demonstrated that increasing lumen occlusion produces sharper luminal pressure peaks when luminal contents exhibit viscosities greater than water, which are expected to generate transient increases in fluid shear. These observations suggest that conventional constant flow conditions or agitation methods may not adequately represent the dynamic and heterogeneous hydrodynamic environment of the human colon.

While none of the systems adequately reproduced dissolution profiles measured under the DCM baseline condition, the USP IV generally demonstrated the closest overall agreement with the maltose-stimulated DCM data, while lower agitation conditions in USP II and TIM-2 also showed comparable profile shapes. Where f₁ values exceeded the recommended threshold despite acceptable f₂ values, this likely reflects the greater sensitivity of f₁ to isolated time-point deviations. Consequently, f₂ is considered the more informative metric for modified-release formulations and is the preferred regulatory measure for dissolution profile comparison.²⁴

Inclusion of dewatering provided additional mechanistic insight into the dissolution experiments, decreasing the rate of dissolution in all systems and improving alignment with the data from the maltose-stimulated DCM (Table 5). In USP IV, the effect was flow rate-dependent: at 25 mL min⁻¹, dewatering substantially improved both f₁ and f₂ relative to the maltose-stimulated DCM, while at 4 mL min⁻¹, where profiles were already closely matched, the effect was marginal. This suggests that the increased medium viscosity achieved through dewatering brings higher flow-rate conditions into closer alignment with the DCM environment by partially offsetting excess fluid shear, but offers diminishing returns where conditions are already well matched.

It should be reiterated that TIM-2 was operated in a simplified configuration in this study, using a simple buffer in place of typical TIM colonic fluid, without mimicked colonic secretion or absorption (filtration), in order to isolate the contribution of hydrodynamics. The multiphase nature of typical TIM colonic fluid (TCF), including the presence of fibre and its associated viscosity, would be expected to significantly alter the hydrodynamic environment within the system and may also influence polymer matrix hydration directly.

Several directions for future investigation emerge from these findings. Pre-conditioning to replicate exposure to the upper GI tract conditions using TIM-1 or tiny-TIM could provide more physiologically relevant matrix hydration prior to colonic testing. Future TIM-2 studies should investigate motility patterns more informed by in vivo MRI and manometry, incorporating periods of quiescence and transient contractile events, as well as operation under standard conditions with active secretion and absorption functions. Studies using higher-dose formulations would help assess the influence of sink limitations and hydrodynamic sensitivity. Finally, integration of dissolution data with validated PBPK models could establish which hydrodynamic conditions provide the greatest biopredictive value and where f₁ and f₂ differences actually translate to clinically meaningful similarity or dissimilarity.

Collectively, these findings suggest that colonic dissolution of hydrophilic matrix tablets is governed by the interaction of matrix hydration, erosion, transient shear, and fluid removal processes. Biorelevant dissolution systems capable of reproducing these interacting phenomena are therefore likely to provide greater predictive value than conventional constant-agitation methods when evaluating colon-targeted modified-release formulations.

4.1. Colonic water absorption

To give physiological context to the water removed from the dissolution medium in vitro during the dewatering experiments, Fig. 7 presents all published data (known to the authors at the time of writing) measuring water absorption from the human colon with the in vitro settings overlayed.^25–29 Whilst there is limited data available, these human data lie across a broad range. This appears to be largely attributable to the test method, since studies quoting data upwards of 50 mL h⁻¹ involved perfusion to push the absorptive capacity of the colon and measure maximum capacity. Whilst the DCM was used for a benchmark comparison in this study, the DCM represents the proximal colon only, therefore it should be expected that its absorption rates are lower than observed in vivo. DCM water absorption rates varied from 5–10 mL h⁻¹ (ref. 3) from sampling and replenishment of the lumen fluid alone. The TIM-2 is designed to represent the entire colon though it has a downscaled volume (approximately 1/3) so default absorption rates are also downsized accordingly, compared to the colon. In the standard TIM-2 settings, absorption rates range from 7.5–60 mL h⁻¹ throughout an experiment following the transient nature of mimicked colonic inflow following a TIM-1 upper GI tract experiment.

Fig. 7. Water absorption rates measured in the human colon and mimicked in vitro. The DCM block represents DCM fluid turnover due to sampling, and the TIM-2 (manual) dashed line represents the values used in this study by manually removing and replenishing lumen fluid; the same rate was applied in both the USP systems in this study.^{25–27,30–32}.

The manual dewatering methodology used in this study in the USP systems and the TIM-2 (50 mL once per hour for the first 3 h plus 1 mL sample volume) was close to the mean absorption rate for the non-perfused human colon but still relatively low (although no standard deviation or range were available for these data).

Owing to the composition of the in vitro dissolution media used in this work and the reference DCM study, dewatering refers to removal of dissolved API and an increase in bulk viscosity. While increased bulk viscosity is expected to retard theophylline dissolution by slowing diffusion and convective mixing, in vivo water absorption removes free water and depletes the liquid volume fraction of the total contents. This mechanistic difference may influence dissolution behaviour in a distinct manner, and likely to differing extents for drugs with various aqueous solubilities, warranting further investigation.

At present, there are no standardised guidelines defining either the duration of dissolution testing incorporating dewatering or the viscosity conditions that should be applied to simulate the colon. However, this would be an interesting future study to explore using advanced dissolution apparatus.

5. Conclusion

This study demonstrates that hydrodynamic conditions strongly influence the dissolution of Uniphyllin® modified-release tablets across all test systems. While USP II and USP IV remain widely used for quality control, their ability to replicate colonic conditions is limited. USP II showed complete release at all paddle speeds within 24 h, but similarity to the physiologically validated DCM was only achieved at the lowest agitation (50 rpm), highlighting the need for reduced shear to mimic colonic motility. USP IV provided more physiologically relevant hydrodynamics, particularly at low flow rates (4–16 mL min⁻¹), which achieved the highest similarity factors (f₂ > 80) against DCM stimulated profiles.

A simplified TIM-2 setup was employed for dissolution testing for the first time and demonstrated motility-dependent release, with slow mixing producing the closest match to DCM (f₂ = 51.8). This confirms that dynamic motility patterns significantly affect drug release and supports TIM-2 as a promising surrogate for colonic dissolution studies. Incorporating dewatering improved similarity across all systems, particularly under stimulated conditions, suggesting that viscosity changes are critical for physiological relevance. However, none of the models fully replicated the DCM baseline profile, emphasizing the complexity of colonic conditions and the need for further refinement of in vitro methods.

Overall, USP methods remain practical for regulatory purposes, but advanced dynamic systems such as TIM-2 and DCM offer potential to integrate more realistic motility patterns to explore colonic drug delivery in vitro. Future work should focus on integrating realistic motility patterns and fluid dynamics into standardized dissolution protocols to bridge the gap between conventional and biorelevant testing.

Author contributions

K. K.: investigation; validation; writing – original draft. E. J.: methodology; writing – review & editing. C. O. F.: methodology; writing – review & editing; H. K. B.: conceptualisation; writing – review & editing; supervision.

Conflicts of interest

Connor O'Farrell is an employee of InnoGI, the company that develops and commercializes the TIM-2 technology used in this study. Their involvement in the research included manuscript review and generation of Fig. 6 on colonic fluid absorption. The use of TIM-2 in this publication may therefore be perceived as a potential conflict of interest. All other authors declare that they have no competing financial or personal interests that could have influenced the work reported in this manuscript. The study was conducted objectively, and all results are presented without bias.

Data availability

Data for this article, including all raw dissolution data are available via the University of Strathclyde's institutional repository, PURE, and can be accessed at the following https://doi.org/10.15129/8dd67a58-1cd8-45a9-924f-b75793d814b1.

The supplementary information contains the power law model parameters fitted to the theophylline dissolution data obtained using USP Apparatus 2, USP Apparatus 4, and TIM-2. Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d6pm00090h.