Development and application of a multi-sugar assay to assess intestinal permeability

Abstract

Aim: Bioanalytical assays to measure rhamnose, erythritol, lactulose and sucralose in human urine and plasma were developed to support an indomethacin challenge study for intestinal permeability assessment in healthy participants.

Methods: The multi-sugar assays utilized 5-μl sample matrix and a simple chemical derivatization with acetic anhydride, followed by RPLC-MS/MS detection.

Results: Rhamnose and erythritol quantification was established between 1.00–1,000 μg/ml in urine and 250–250,000 ng/ml in plasma. For lactulose and sucralose, dynamic ranges of 0.1–100 μg/ml (urine) and 25–25,000 ng/ml (plasma) were applied for biological measurements.

Conclusion: This work helped overcome some of the common analytical challenges associated with the bioanalysis of mono- and disaccharides and achieved improved limits of quantification.

Plain language summary

Article highlights.

Background

• Permeability ratios of lactulose to rhamnose, as well as sucralose to erythritol are important biomarkers for assessing intestinal permeability modulation.

• Accurate quantification of lactulose, rhamnose, sucralose and erythritol in both urine and plasma was achieved using a simplified sample procedure and RPLC-MS/MS detection.

Methods

• The direct addition of acetic anhydride and pyridine onto 5 μl urine or plasma matrix enabled efficient derivatization of the sugar analytes and straightforward separation by RPLC and MS/MS detection.

• For urine analysis, a 20-min RPLC method was optimized for the four sugar derivatives to ensure selectivity and limit isomer/isobaric interferences, while a 16-min run time was applied for plasma.

• Technical considerations such as MS tune settings and mobile phase selection were important for driving favorable ammonium adduct formation with sugar molecules.

Results & conclusion

• Using the final RPLC-MS/MS methods, the disaccharides – lactulose and sucralose could be measured down to 0.1 μg/ml in urine and 25 ng/ml in plasma using only 5 μl matrix and limited sample handling/enrichment – a vast improvement for this type of method.

• Overall, these reproducible and precise (within 20%) analytical methods enabled robust assessment of intestinal barrier permeability and can help provide confidence in disaccharide to monosaccharide ratio end points with modest changes.

• The concepts of this RPLC-MS/MS work may be broadly applied to other bioanalysis requests involving saccharides.

1. Background

The intestine is designed to be semi-permeable by acting as a barrier against bacteria and other infectious agents while allowing water and other vital nutrients from our food to pass into the bloodstream. However, individuals with inflammatory bowel diseases often have increased intestinal barrier permeability (otherwise known as ‘leaky gut’), which is hypothesized to contribute to their unsought symptoms [1]. Treatments are being developed that focus on improving intestinal barrier permeability [1–10]. Akin to that, improved analytical tools to accurately measure small changes in intestinal permeability in vivo are critical, especially for clinical studies with interventional treatment [1–3,11–15]. A common way to assess human intestinal permeability is by measuring urinary excretion concentrations of orally administered sugar probes such as a mix of disaccharides (e.g., lactulose, sucralose) and monosaccharides (e.g., rhamnose, erythritol) [3,12].

Orally ingested sugar probes can cross the intestinal epithelium, enter the bloodstream and then filter through the kidneys without reabsorption back into the blood [2,16]. Urinary and plasma measurements of these sugars are utilized to calculate permeability ratios (e.g., disaccharide/monosaccharide concentrations) to evaluate the integrity of the small and large bowels [1–3,11,12]. Studies have used these permeability ratios to understand gastrointestinal function previously [1,2,11–13] with the notion that the larger disaccharide (lactulose or sucralose) can cross the epithelium at a site of damage or via a ‘leaky pathway’, in contrast to the monosaccharide, which is readily absorbed [1,12]. For instance, van Wjick et al. [12] previously reported a study involving administration of a multi-sugar mixture containing sucrose, lactulose, rhamnose, erythritol and sucralose to assess site-specific intestinal permeability via an isocratic cation exchange LC-MS/MS method [11,12]. Lactulose-to-rhamnose ratios were used to evaluate the small intestine, while sucralose and erythritol urine/plasma concentrations were used to signify permeability along the entire intestinal tract, or more specifically, in the colon when limiting assessment between the 5- and 24-h sampling window [1,2,11,12].

Several groups have reported innovative separation methods via LC-MS/MS to quantify saccharides in urine or plasma matrices [11,13–22]. However, there is still a need to improve assay sensitivity and precision for new applications in clinical studies as saccharide measurement by LC-MS/MS and other bioanalytical methods is a difficult challenge. To start with, multiple chiral centers in some sugar molecules can create a dilemma for bioanalysis because chromatographic isomeric/isobaric resolution is required and a critical aspect for ensuring accurate quantification. For example, rhamnose and fucose have the same molecular mass and formula but differ slightly in structure/shape. The same concept applies for lactulose, lactose and sucrose. Lactulose and lactose have high chemical similarities where lactulose is a lactose derivative formed by chemical isomerization [23]. The takeaway is that checking for chromatographic separation of these isomers is vital to maintain assay selectivity. Keto-enol tautomerism of reducing sugars (i.e., rhamnose and lactulose) can also be a hinderance for LC-MS/MS development because this interconversion can affect chromatographic peak shape, and more than one peak may be observed during separation. A third challenge for this work is that sugar molecules lack ionizable moieties that lead to poor electrospray ionization (ESI) for mass spectrometric detection and adduct formation with metal cations or salts are often observed under ESI conditions. Finally, from a LC-MS/MS perspective, saccharides have high hydrophilic properties which makes reversed-phase LC-MS/MS (RPLC-MS/MS) not ideal without derivatizing the sugar first to retain the molecule on its stationary phase. RPLC-MS/MS is a common set-up in bioanalytical labs though, whereas other separation techniques (e.g., ion exchange or hydrophilic interaction chromatography, etc.) are less typical but still plausible. From a biological matrix standpoint, challenges such as endogenous sugars in urine and plasma usually necessitate that a surrogate matrix is needed for proper bioanalysis. Furthermore, insufficient recovery and concerns of interference during saccharide extraction need to be carefully addressed. Overall, saccharide LC-MS/MS bioanalytical methods can be tedious to develop to ensure high quality data, but this technique can provide confidence, especially with targeted LC-MS/MS detection.

For this paper, a sensitive, selective, and reproducible RPLC-MS/MS method was developed to measure four sugars: lactulose, rhamnose, sucralose and erythritol in human urine/plasma and addresses some of the technical challenges mentioned above. This work was performed as part of a capability assessment to support a non-investigational medicinal product (NIMP) study which applied an indomethacin challenge in healthy participants. Indomethacin is a commercially available non-steroidal anti-inflammatory drug (NSAID) which, when administered to healthy participants causes ulcerations of the small intestine, similar to those seen in patients with Crohn's disease [12,24–26]. The indomethacin challenge study was designed to assess small intestinal permeability using lactulose/rhamnose ratios and colonic permeability using sucralose/erythritol ratios. Healthy participants were dosed with either (oral) placebo or indomethacin, and then dosed with the multi-sugar probe (lactulose, rhamnose, sucralose and erythritol) cocktail. Urine and plasma samples from healthy volunteers were collected over a 24-h period for RPLC-MS/MS bioanalysis. Herein, this work takes a fresh look into ways to overcome some of the analytical and technical challenges for measuring multiple sugars within different matrices from sample preparation and surrogate matrix evaluation to dynamic range determination and RPLC-MS/MS optimization.

2. Materials & methods

2.1. Reagents & materials

The following reference materials; erythritol, rhamnose monohydrate, lactulose, [¹³C₁₂]- lactulose and sucralose; were obtained from Sigma-Aldrich (St. Louis, MO, USA, >98%). [²H₆]-Erythritol and [²H₆]-sucralose were purchased from Toronto Research Chemicals (Toronto, ON, Canada). [¹³C₆]-Rhamnose monohydrate was purchased from Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA). Laboratory solvents and reagents such as acetonitrile, isopropanol, acetic anhydride, formic acid, ammonium bicarbonate and methanol were of LC/MS grade and purchased from Sigma-Aldrich (St. Louis, MO, USA). Pyridine, >99.0%, was purchased from Thermo Fisher Scientific (Waltham, MA, USA).

2.2. Preparation of calibration standards & quality control samples

Stock solutions of the investigated sugar probes and their respectively labeled internal standards (IS) were prepared by dissolving reference standards in water and kept at 4°C. The final concentration of individual sugar was 250 mg/ml for erythritol, 25 mg/ml for rhamnose and 5 mg/ml for lactulose and sucralose. The concentration of the IS was 10 mg/ml for [¹³C₆]-rhamnose and [¹³C₁₂]- lactulose while [²H₆]-erythritol and [²H₆]-sucralose were prepared at 1 mg/ml.

For the urine assay, working solutions containing a mixture of all four analytes were made from the stock solutions at 5000, 500 and 50.0 μg/ml for rhamnose and erythritol and at 500, 50.0 and 5.00 μg/ml for lactulose and sucralose. The working solutions were made fresh in human urine dialysate on the day of analysis and used to prepare eight calibration standards in dialyzed human urine over the range of 1.00–1,000 μg/ml for rhamnose and erythritol and 0.1–100 μg/ml for lactulose and sucralose.

For the plasma assay, working solutions containing a mixture of all four analytes were made from the stock solutions at 1250, 125 and 12.5 μg/ml for rhamnose and erythritol and at 125, 12.5 and 1.25 μg/ml for lactulose and sucralose. The working solutions were made fresh in human plasma dialysate on the day of analysis and used to prepare eight calibration standards in dialyzed human plasma over the range of 250–250,000 ng/ml for the monosaccharides (rhamnose and erythritol) and 25–25,000 ng/ml for the disaccharides (lactulose and sucralose).

2.3. Dialysis of human urine or plasma to remove endogenous sugars

To generate control (surrogate) matrix, pooled human urine or plasma were separately dialyzed to remove endogenous sugars. Specifically, 10–40 ml of human urine or plasma were added to benzoylated dialysis tubing (Millipore Sigma, cat no: D2272-10FT or D7884-5FT) and enclosed using dialysis tubing clamps (Millipore Sigma, cat: Z371092). The dialysis tubes containing urine or plasma were then immersed into a 2 L glass beaker containing water for approximately 4 h with constant stirring. The water was then replaced with fresh water for another round of dialysis overnight. The next day, the dialyzed urine or plasma was removed from the tubing, aliquoted and stored at -20°C until use. The dialyzed human urine or plasma were tested as blank samples to ensure relevant endogenous sugars were removed prior to use for calibration standard and quality control (QC) preparation.

2.4. Extraction of the evaluated sugar probes from human urine or plasma

An aliquot of 5 μl of human urine or plasma from incurred, calibration standards, QC and control blanks samples was transferred into 1.4 ml micronic tubes (MP32022; Thermo Fisher Scientific, USA). An aliquot of 150 μl pyridine was added to double blank samples while the rest of the samples received 150 μl pyridine containing [²H₆]-erythritol, [¹³C₆]-rhamnose, [¹³C₁₂]- lactulose and [²H₆]-sucralose as IS. The concentration of the IS was 0.5 and 0.125 μg/ml for the urine and plasma assays, respectively. Then, 50 μl acetic anhydride was added to all samples to start the derivatization reaction. The acetic anhydride stock was stored in the fridge and was left to acclimatize at ambient condition for at least 20 min before use. After vortex-mixing for approximately 1 min, the samples were incubated at 37°C for approximately 1 h under constant shaking. After completion of the reaction, the samples were centrifuged at 3000×g for 3 min. An aliquot of 100 μl of the reaction mixture was transferred to a fresh 1.4 ml polypropylene micronic tube using a TomTec Quadra 3 liquid handler (Tomtec, CT, USA), and the samples were dried down under a steady stream of nitrogen at 45°C. Finally, the sample extracts were reconstituted with 200 μl 30% acetonitrile in water followed by analysis via RPLC-MS/MS.

2.5. Chromatographic conditions for the analysis of human urine extracts

Derivatives from urine extracts (1 μl using a partial loop mode) were injected onto an Ultra Performance Liquid Chromatography (UPLC) system (Waters, Milford, MA, USA) using a 2.1 mm × 150 mm Halo C18, 90 Å, 2.7 μm particle size analytical column (Advanced Materials Technology, Wilmington, DE, USA). Mobile phase A consisted of 50 mM ammonium bicarbonate (native pH), and mobile phase B was a mixture of acetonitrile and methanol 80:20 (v/v). The column was kept at ambient temperature during the entire run. The chromatographic separation was performed using an isocratic mode at 35% B until 5 min followed by another isocratic mode at 40% B from 5.1 to 15 min. Then, a linear gradient from 70 to 85% B between 15.1 and 17.3 min was applied, followed by an isocratic profile at 35% B from 17.4 to 20 min. The flow rate was maintained at 0.25 ml/min during the entire run. A mixture of acetonitrile and water (70:30 v/v; 1000 μl) was used as weak wash while acetonitrile/isopropanol/water/formic acid (40/40/20/0.1 (v/v/v/v), 1,000 μl) was used as the strong wash.

2.6. Chromatographic conditions for the analysis of human plasma extracts

Derivatives from plasma extracts were analyzed using the same UPLC system, analytical column, flow rate and mobile phase composition as for the human urine assay, except changes to the gradient profile were applied. The run time was shortened by 4 min compared with the urine assay. Specifically, the plasma extracts were analyzed using isocratic mode at 30% B until 3 min, followed by a linear gradient from 30 to 38% B ending at 5 min. Then, another isocratic mode at 38% B was used until 12.5 min, followed by two short gradients from 38 to 70% B (12.5–13.5 min) and 70% to 85 (13.5–15 min). The system was then equilibrated at 30% B for a minute before the next injection.

2.7. Mass spectrometric conditions

A tandem triple quadrupole mass spectrometer Xevo TQ-XS (Waters, Milford, MA, USA) was operated in the electrospray (ESI) positive ionization mode. Single reaction monitoring (SRM) of the acetyl derivatives were used to ensure the required selectivity. The optimized instrument parameters included: source temperature, 400°C; capillary voltage, 2.0 kV; cone voltage, 10 V; nebulizer, 7 Bar; and declustering potential, 65 eV. Information on the SRM transitions for each derivatized sugar analyte is detailed later.

2.8. Indomethacin challenge study & sampling from healthy human volunteers

Details of the indomethacin challenge methodology are described elsewhere [12]. The study in healthy participants was a NIMP study. The study protocol, informed consent and other information that required pre-approval were reviewed and approved by the London-Riverside Research Ethics Committee (REC) in accordance with the ICH GCP and applicable country-specific requirements. In summary, participants took either indomethacin or placebo using a fixed clinically approved dosing regimen as used in previous studies. Specifically, urine/plasma (pre-dose) samples were collected prior to giving each participant the sugar mixture for ingestion. The dissolved sugar mixture contained sucralose, erythritol, rhamnose and lactulose at 0.5 g rhamnose and 1.0 g for all other components, in 150 ml water. In addition to the pre-dose sample, urine samples were collected on Day 1 between 0–2 h (2 h), 2–5 h (5 h) and 5–24 h (24 h). De-identified urine samples were frozen immediately at -20°C or lower until analysis. Blood samples were collected in K₂EDTA vacutainers at the following time-points: pre-sugar mix (0 h) and after the administration of the sugar mix at: 1-, 2-, 3-, 4-, 5-, 8- and 24-h. Blood samples were then centrifuged at 1,500 rcf (g) at 2–8°C for 15 min and plasma samples harvested and stored at -20°C or lower until analysis. The bioanalysis was blinded to treatment type (indomethacin vs. placebo administration) during sample testing.

3. Results & discussion

The analysis of saccharides in biological matrices has its challenges, especially in terms of poor matrix recovery, chromatographic retention, assay selectivity and MS detection. In 2016, our laboratory published [14] a relatively simple derivatization procedure with acetic anhydride to quantify four sugar probes (mannitol, lactulose, sucrose and sucralose) down to 0.5 μg/ml in cynomolgus plasma and urine. Some of those key learnings were leveraged for this current work and adapted. However, the goal of the current work was to confidently quantify sugar probes: erythritol, rhamnose, lactulose and sucralose in both human urine and plasma to ultimately support an intestinal permeability challenge study where these sugar probes were administered to healthy volunteers following acute indomethacin administration. In addition, there was an aim to lower the quantification limit closer to 0.1 μg/ml in human urine and 25 ng/ml in plasma for the disaccharides: lactulose and sucralose.

Several key technical points are discussed herein that expands past our previous 2016 work [14] in order to develop a reliable multi-sugar LC-MS/MS assay with an improved LLQ for the disaccharides. For example, attention to LC-MS/MS tune parameters and its impact on adduct formation was an important learning. Diligence to confirm the selectivity of the investigated sugars was also necessary, especially for the sugar analytes that may have interference from structurally related stereoisomers. Extensive efforts were spent during method development to ensure that rhamnose selectivity was not impacted by fucose and other isobaric endogenous molecules. Similar efforts were dedicated to confirming that the presence of lactose and its related impurities would not impact the selectivity of lactulose. Moreover, deciphering the appropriate dynamic ranges for each sugar probe in either urine or plasma was required. Luckily, some previous empirical data from a similar NSAID challenge using these sugars as probes [11,12] could be leveraged when selecting the dynamic linear ranges. Tuning or de-tuning SRM parameters for certain sugars was another critical aspect for the final assays. In summary, this new work highlights fundamental points to consider for developing high-quality LC-MS/MS methods for multiple sugar analytes in urine and plasma with improved LLQs, especially for lactulose and sucralose. The final assays were then applied to support an indomethacin challenge study, and some representative data of the individual sugar concentrations in urine and plasma are presented for proof-of-concept.

3.1. Modifications made to the methodology reported by Licea-Perez et. al., 2016

The analysis of sugar probes was previously performed after derivatization with acetic anhydride, as described in the Licea-Perez et. al. 2016 paper [14]; however, some modifications to improve upon the method (i.e., lower LLQ) were introduced. First off, the disaccharides, lactulose and sucralose were kept in the assay while mannitol was replaced with two monosaccharides (erythritol and rhamnose) to fit current study needs. The chemical structures of erythritol, rhamnose and their respective acetic acid derivatives are shown in Figure 1. Second, the matrix volume was reduced to 5 μl instead of 10 μl, which enabled the derivatization step to be performed directly on the sample (urine or plasma) without prior precipitation of plasma proteins or sample dry-down. These modifications were implemented to improve assay precision and most importantly analyte recovery, so that the quantification limit could be lowered fivefold (0.1 μg/ml and 25 ng/ml in urine and plasma, respectively) for lactulose and sucralose as compared with that for the previously reported assay [14]. These modifications to the sample preparation also improved assay efficiency by limiting the number of steps in the protocol. Improving the LLQ was necessary especially for the lactulose analyte as its urinary / plasma recovery at baseline (e.g., pre-dose samples) has been reported to be close to the published LLQ in previous LC-MS/MS assays focused on intestinal permeability measurement [1,11].

Figure 1.

Scheme for the derivatization reaction between rhamnose and erythritol with acetic anhydride.

3.2. Investigation into the adduct formation on the LC-MS/MS analysis

It is well known that carbohydrates/saccharides have poor ionization efficiency using conventional MS ionization techniques. Ammonium adducts of the acetic acid derivatives were successfully used in the bioanalysis of mannitol, lactulose, sucrose and sucralose in our previous publication that utilized a Sciex mass spectrometer [14]. To improve the sensitivity of the disaccharides (lactulose and sucralose) in question, MS parameters on a Waters TQ-XS mass spectrometer were optimized to drive ammonium adduct formation instead of sodium adducts. The ammonium adducts were significantly enhanced by adding (1) ammonium bicarbonate in the mobile phase (50 mM, native pH) and (2) reducing the desolvation temperature in the ion source to 400°C. The impact of the desolvation temperature on the adduct formation is illustrated in Figure 2 where higher values were in favor of the formation of sodium adducts (Figure 2C & F) while low temperatures benefited the formation of ammonium adducts (Figure 2A & D). When [M+Na]⁺ adducts formed, product ions were difficult to generate even with high collision energy applied. Ammonium adducts, [M+NH₄]⁺, however, produced reasonable SRM transitions for each of the sugar derivatives. These observations also agreed with the findings reported by Keane et al. using Sciex MS systems [27] and demonstrate that this concept should be universal to different MS systems.

Figure 2.

Impact of the desolvation temperature on the adduct formation during ESI-MS/MS for lactulose and rhamnose derivatives. (A–C) represent the spectra for lactulose derivatives at 400, 500 and 650°C, respectively. (D–F) represent the spectra for rhamnose derivatives at 400, 500 and 650°C, respectively.

3.3. Addressing selectivity challenges for rhamnose & lactulose

SRM method development for both erythritol and sucralose was mainly straightforward; however, rhamnose and lactulose had some challenges which needed to be addressed to increase confidence in the multi-sugar RPLC-MS/MS assay. Specifically, both rhamnose and lactulose are reducing sugars that mean that these molecules may exist in two distinct forms, the open (linear) and closed-ring form. These two forms often have different physicochemical properties, resulting in two distinct chromatographic peaks. For rhamnose, the derivatization with acetic anhydride consistently generated two equally intense chromatographic peaks at retention times of 7.5 and 9.8 min. For lactulose, two chromatographic peaks were observed at 11.3 and 13.5 min; however, the intensity of the later eluting peak was predominant (95% of the total response). For both rhamnose and lactulose, the second eluting peak was selected for quantitation. Identification of the forms and corresponding peaks for the reducing sugars was beyond the scope of this work and thus not further investigated.

Interference in the SRM channels was another challenge that needed to be addressed. For example, the SRM chromatogram for rhamnose after extraction from human urine was very congested and contained several peaks that contributed from the biological matrix itself. This was observed mainly for the first eluting rhamnose peak at 7.5 min as it was among several high-intensity endogenous peaks. Fortunately, no endogenous peaks were observed for the rhamnose SRM peak at 9.8 min, and thus was selected for quantification purposes. Two contributing factors for the adequate separation and selectivity of the rhamnose peak at 9.8 min were due to the relatively long analytical column (2.1 mm × 150 mm Halo C18, 90Å, 2.7 μm particle size) and the use of a shallow gradient. To give further confidence in the selectivity of the urine assay, urine extracts were analyzed using a 20-, 30- and 60-min gradient and the peak area ratios of each sugar analyte-to-IS were compared. The intention of this method development experiment was to show that the chromatographic peak at 9.8 min corresponded to the rhamnose derivative and without any potential interference from other endogenous components within the same SRM channel. As there was no significant difference in the peak area ratios and no extra peaks observed in the 30- nor 60-min RPLC methods, this experiment helped ensure that the 20-min method provided the appropriate separation and selectivity for rhamnose after extraction from human urine. Furthermore, no interference was observed from the stereoisomer fucose which has the same molecular weight and SRM as rhamnose (Figure 3A & B). Finally, several commercially available (pooled) human urine lots were analyzed using the 20- and 60-min RPLC methods and the results were comparable indicating that the 20-min method was selective for the analysis of rhamnose.

Figure 3.

Interference screening for the multisugar assay. (A) represents a chromatogram of extracted human urine lysate containing 5,000 ng/ml spiked fucose while (B) is a sample extract containing both fucose and rhamnose spiked at 5,000 ng/ml. Note: the scheduled SRM channel for rhamnose starts at 9.0 min. (C & D) demonstrate interference screening using 5,000 ng/ml (spiked) lactose and its related impurities monitored via the SRM transition of lactulose. (D) was made by zooming (C) in by two orders of magnitude.

Assay selectivity for lactulose also needed to be addressed as this synthetic sugar is an isomer of lactose. A selectivity check for the lactulose assay indicated that the main hurdle was associated with the presence of very high background levels of lactose in some of the investigated pooled human urine lots. The derivatization of (matrix-free) lactose reference standard showed that its main derivative after acetic anhydride treatment contained eight acetylated hydroxyl moieties in comparison to seven for lactulose. Specifically, the main derivative of lactose had a higher mass (SRM m/z 696 to 211) versus lactulose (SRM m/z 654 to 211) which contained seven acetylated hydroxyl moieties. This phenomenon was also consistently observed in the previously reported 2016 method [14]. Furthermore, the retention time of the main lactose derivative differed compared with the lactulose derivative (16.7 vs. 13.5 min, Figure 3C & D). However, there were two low-abundant interference peaks in the lactulose SRM transition (m/z 654 to 211) that eluted near its expected retention time when a high concentration of pure lactose standard (5,000 ng/ml) was spiked-in and derivatized (Figure 3D). These near-by interference peaks were observed somewhat in some human urine lots analyzed during the selectivity screening. To ensure a highly selectivity method, this was the main reason at least a 20-min RPLC run time was needed. Overall, these interference peaks nearby the expected retention time of the lactulose derivative were of low abundance, deeming any potential interference to be low-risk. However, a dietary restriction on lactose was necessary throughout the NIMP study. On a similar note, selecting a lactulose standard or dosing solution with high purity was critical to avoid lactose contamination.

A multi-sugar assay for human plasma was also developed in parallel to have flexibility around the experimental testing strategy. Similar experiments were performed as described above to confirm the selectivity for rhamnose and lactulose in different human plasma lots. The SRM transitions of rhamnose and lactulose derivatives were cleaner in the plasma extracts when compared with urine, and therefore, the RPLC run time was reduced to 16 min. The composition of plasma and urine differ with plasma mainly containing proteins, such as IgG and albumin. Albeit these larger proteins in plasma, column clogging was not observed to be an issue for the assay since only 5 μl of plasma was used. Ultimately, the RPLC-MS/MS background for the plasma assay was expected to be cleaner due to less sugar /small molecule background levels, and thus a shorter run time was possible for improved bioanalysis efficiency.

3.4. Selection of the dynamic range of the analytes in human urine & plasma

Another point to consider was the endogenous background levels of these sugars in human urine and plasma. Previous literature data was leveraged to help predict expected sugar levels in urine/plasma at baseline and after dosing. Particularly, levels of erythritol and rhamnose in human urine and plasma were reported to be relatively high [11]. In our lab, several commercially available lots of human urine and plasma were analyzed using our RPLC-MS/MS assay, and it was confirmed that erythritol and rhamnose endogenous concentrations were high, specifically in the μg/ml range. Therefore, dynamic ranges for rhamnose and erythritol of 1.00–1,000 μg/ml and 0.250–250 μg/ml were chosen for the urine and plasma assay, respectively. On a similar note, dynamic ranges of 0.1–100 μg/ml (urine) and 25–25,000 ng/ml (plasma) for the lactulose and sucralose bio-assays were selected. Representative chromatograms of the extracted analyte derivatives at the LLQ from human urine and plasma dialysate are shown in Figure 4 (urine) & Supplementary Figure S1 (plasma), respectively.

Figure 4.

Extracted ion chromatograms from human urine dialysate at the LOQ level for (A) erythritol, (B) rhamnose, (C) lactulose and (D) sucralose. The SRM transitions for erythritol and rhamnose correspond to the ¹³C isotope.

With endogenous sugars (i.e., erythritol, rhamnose and lactose) in some human urine /plasma pooled lots, a surrogate matrix was obviously needed for proper sugar analyte quantification and to prepare calibration and QC standards. At first, both UriSub^® (for urine) and SeraSub^® (for plasma) (CST Technologies, NY, USA) were evaluated as surrogate matrices to prepare calibration standards and QC samples; however, these matrices were abandoned as the IS response was slightly higher than compared with authentic matrix/samples. This issue was not observed though when dialyzed urine and plasma were utilized. To that end, dialyzed matrix was selected for the assay. The removal of endogenous analytes was performed using benzoylated dialysis tubing in a 2-L glass beaker containing water under constant stirring, and the water was replaced once over 24 h. Figure 5 shows representative chromatograms of erythritol and rhamnose acetic acid derivatives before and after the urine dialysis procedure in which less than ≈1% of endogenous sugar remained.

Figure 5.

Representative SRM chromatograms of authentic human urine before (A) erythritol, (C) rhamnose and after (B) erythritol, (D) rhamnose dialysis.

As the dynamic ranges of the investigated di- and monosaccharides were quite different, instrument parameters were optimized individually for each sugar and a scheduled SRM method workflow was applied. For example, SRM signals were enhanced for lactulose and sucralose derivatives whereas a de-tuning approach was applied for erythritol and rhamnose derivatives. To further elaborate, the MS instrument parameters were thoroughly optimized for lactulose and sucralose acetic acid derivatives to achieve a new LLQ. The highest impact on the sensitivity and formation of ammonium adducts with the derivatized sugars happened when the desolvation temperature in the ion source was set to 400°C. To achieve relevant dynamic ranges for the monosaccharides though, de-optimized SRM transitions corresponding to the ¹³C isotope were selected in the urine assay which helped maintain the instrument response linearity. With a total of eight SRM transitions (i.e., four sugars with & without IS) for this multi-analyte assay, a scheduled SRM method with no more than two transitions at a time was wise to improve assay variability and sensitivity where needed. To summarize, the masses of the investigated sugar probes, along with the ammonium adduct masses of their acetyl derivatives are shown in Supplementary Table S1. The SRM transitions used for quantitation of the investigated sugar derivatives for the urine and plasma assays are also illustrated in Table 1 & Supplementary Table S2, respectively.

Table 1.

SRM transitions used for quantitation of the investigated sugar derivatives in the urine assay.

Analyte	Precursor Ion (m/z)	Product Ion (m/z)	Dwell Time (msec)	Scheduled SRM	Collision Energy (eV)
Erythritol	309.0†	130.0	150	Start: 5 min End: 8 min	23
[2H6]-Erythritol	314.0	135.0			20
Rhamnose	351.0†	154.0		Start: 9 min End: 11 min	24
[13C6]-Rhamnose	356.1	159.0			20
Sucralose	624.0	307.0		Start: 15 min End: 19 min	22
[2H6]-sucralose	630.0	309.0			18
Lactulose	654.4	211.0		Start: 12 min End: 15 min	18
[13C12]- lactulose	666.4	217.0			20

m/z represents ammonium adduct of acetyl derivatives.

^†m/z represents ammonium adduct of acetylated derivatives of the ¹³C isotope (detuned SRM transition).

3.5. Assay qualification & application to a NIMP study

After completing method development, the assays were qualified by performing three precision and accuracy runs in both urine and plasma dialysates. The runs passed pre-defined acceptance criteria for the percent bias and percent coefficient of variation (%CV) of ≤ ± 20. Also, eight different lots of commercial human urine and plasma (BioIVT, NY, USA) were analyzed in each run and the within run-to-run variability was deemed acceptable (% CV of ≤ ± 20).

After performing the qualification runs, the two multi-sugar assays (for urine and plasma) were considered suitable for monitoring the sugar probes for the NIMP study. A total of 228 urine and 456 plasma samples were analyzed in a total of four (urine) and seven (plasma) analytical runs. The QC statistics from the bioanalysis study are plotted in Figure 6 (for urine) & Supplementary Figure S2 (for plasma) for each of the individual sugar analytes. The runs met the pre-defined acceptance criteria in which three-fourths of the QC sample results at each level measured within 20% of the nominal concentrations. Overall, these new RPLC-MS/MS assays were conducted to enable accurate measurement of two monosaccharides and two disaccharides in both urine and plasma. For method demonstration purposes, concentration-time profiles for each sugar probe from a single, representative sample are shown in Figure 7 for the urine assay. This example profile is not intended to inform intestinal permeability (out of scope for this analytical-focused paper) but rather used to illustrate that this revamped bioanalytical assay could measure the sugar probes within the intended urinary excretion/concentration range expected for either indomethacin or placebo treated subjects, deeming it fit for purpose. As for the plasma bioanalysis, a representative concentration-time profile of a (de-identified) subject administered with either indomethacin or placebo is also shown in Supplementary Figure S3 and demonstrated that plasma analysis may also add value for intestinal permeability understanding.

Figure 6.

Summary of the assay performance during the sample analysis of the investigated sugar probes (A) erythritol, (B) rhamnose, (C) sucralose, (D) lactulose in human urine dialysate, expressed as the percent difference between calculated vs. nominal concentration (n = 16 across four runs).

Figure 7.

A representative concentration-time profile for each sugar probe (A) erythritol, (B) rhamnose, (C) sucralose, (D) lactulose in human urine after either indomethacin or placebo administered.

Overall, this work highlights how >600 samples were analyzed to generate multiple sugar probe concentration read-outs using only two bio-assays with technical points discussed. Improving assay LLQ for lactulose and sucralose enabled more concentration read-outs in urine and plasma, especially for the disaccharides which is critical to enable ratio reporting and for intestinal permeability interpretation. These robust methods may be more practical of use for future clinical studies and/or applied more broadly for other saccharide bioanalyses.

4. Conclusion

In summary, two multi-sugar bioanalytical assays were successfully developed and employed to analyze four different sugar probes in human urine or plasma using a simple sample handling protocol and traditional RPLC-MS/MS methodology. These methods can be used to assess intestinal permeability in either urine and/or plasma from either healthy volunteers or patients with gastrointestinal disease in future interventional studies assessing barrier function, and allows flexibility in testing strategy (e.g., urine or plasma sampling). For the sample preparation, the hydroxyl moieties on the sugar molecules were derivatized with acetic anhydride which increased their RPLC chromatographic retention and enhanced ESI-MS ionization via ammonium adduct formation. Overall, this work successfully supported an exploratory assessment of intestinal permeability in healthy volunteers whose barrier function was perturbed by acute ingestion of an NSAID. The extraction protocol and the MS detection were optimized to achieve a new LLQ of 0.1 μg/ml for lactulose and sucralose in urine. The endogenous levels of erythritol and rhamnose in human urine were relatively high; thus, de-optimized detection parameters were used to maintain assay linearity over the range of 1.00–1000 μg/ml. As lower levels were expected in human plasma, the dynamic ranges were designed to be 25–2500 ng/ml and 0.25–250 μg/ml for the monosaccharides and disaccharides, respectively. Lastly, the general concepts from this work can be easily modified for the bioanalytical LC-MS/MS development of other mono- and disaccharides.

Supplemental material

Supplemental data for this article can be accessed at https://doi.org/10.1080/17576180.2024.2374168

Financial disclosure

The authors have no financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Competing interests disclosure

All authors were employees of GSK during the course of this work. N Schneck, C Teague and HL Perez are currently shareholders. The authors have no other competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript apart from those disclosed.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.

Ethical conduct of research

The authors state that they have obtained appropriate institutional review board approval for all human investigations. The study protocol (GSK Study Register 218678), informed consent and other information that required pre-approval were reviewed and approved by the London-Riverside Research Ethics Committee (REC) in accordance with the ICH GCP and applicable country-specific requirements. In addition, written informed consent has been obtained from all the participants involved.