Lactulose

Environmental enteropathy (EE) is a syndrome of altered small intestine structure and function described among developing world populations with high exposure to enteric pathogens. Characterized by partial villus atrophy and crypt hyperplasia 1, EE has been associated with reduced absorptive capacity, increased permeability 2, and chronic intestinal inflammation 3. EE is hypothesized to underlie the negative synergistic relation between undernutrition and enteric infections observed among children living in poverty 4, as well as to explain oral vaccine failure and the relatively disappointing results of both nutritional interventions and diarrheal disease control measures in improving child growth.

The lactulose:mannitol (L:M) test has been the most popular dual-sugar test used to assess intestinal barrier function for more than 3 decades 5-11. The absorption of the monosaccharide mannitol and the disaccharide lactulose are often said to reflect transcellular and paracellular permeability, respectively, such that mucosal cell damage and damaged tight junctions are associated with decreased mannitol absorption and increased lactulose absorption 12. Although this model remains controversial 11, the L:M test is of evident usefulness in identifying malabsorption and altered permeability among patients experiencing disorders such as celiac disease and inflammatory bowel disease 13, and is the principal diagnostic test of altered gut integrity in the context of studies aiming to quantify the causes and consequences of EE. For instance, it has been used in studies relating the effect of nutrition or sanitation-based interventions to the gastrointestinal health of the child living in poverty 14-20, and several reports have found the L:M ratio (or the excretion ratios of mannitol or lactulose individually) to be associated with growth faltering among at-risk infants 2, 21, 22.

Despite the test's promise, the utility of the L:M ratio to EE research remains constrained. Heterogeneity in the testing protocol, instrumental performance, and reporting does not allow for the comparisons of reported values across studies. Differences in the testing protocol, related to the dosage of sugars given, the solution osmolarity, and the urine collection time, are known to affect urine sugar concentrations and therefore the final L:M ratio 23. Adding to these methodological issues, differences between high-performance liquid chromatography (HPLC) platforms and protocols are recognized to lead to differences in test performance 24, which are different than earlier and classic studies using enzyme-based assays 9, 10, 25. Reported L:M ratios may refer to excretion ratios (the ratio of administered lactulose excreted to administered mannitol excreted), or to concentration ratios (the ratio of the absolute urinary concentration of lactulose to mannitol, wherein concentrations are in milligrams or millimoles per liter).

There are no international reference or measurement standards that facilitate the standardization of the lactulose and mannitol assay. Because L:M values cannot be compared with confidence between laboratories, the point above which a test result is interpreted as “abnormal” is generally defined on a study-by-study or laboratory-by-laboratory basis. Although reference values for small numbers of healthy children have been published 26, to date it has not been possible to reliably relate these values to populations wherein EE is believed to be prevalent. Greater comparability would make it possible to accurately diagnose a child as having an “abnormal” L:M ratio according to a universal, rather than a relative and study-specific standard. This would facilitate epidemiological research by allowing inferences about the comparative prevalence and intensity of EE between populations.

To estimate platform performance over a range of concentrations, we compared absolute urinary concentrations of lactulose, mannitol, and the L:M ratio between 3 independent laboratories: 1 running an high-performance liquid chromatography with pulsed amperometric detection (HPLC-PAD) system and 2 running liquid chromatography tandem mass spectrometry (LC-MSMS) systems, in 100 samples from a cohort of Peruvian infants considered at risk for EE.

METHODS

The Etiology, Risk Factors and Interactions of Enteric Infection and Malnutrition and the Consequences for Child Health (MAL-ED) birth cohort is an ongoing longitudinal multicountry investigation of enteric infection, morbidity, and dietary intake as they affect child development 27. The Peruvian community of Santa Clara de Nanay is 1 of 8 sites participating in this study, which was designed to follow approximately 200 mother–child dyads per site through 24 months of age. Recruitment was ongoing between January 2010 and February 2012. All healthy singleton newborns born to mothers living in the study catchment area, who were at least 16 years old, without another child participating in the study, and who were not planning to travel in the near future were invited to enroll.

Per the MAL-ED protocol, the L:M test was administered to each infant at 3, 6, 9, and 15 months of age. Participants were fasted for a minimum of 2 hours before and 30 minutes following the administration of the disaccharide solution, with the exception of breast milk, which was permitted ad libitum. The solution included 250 mg/mL of lactulose and 50 mg/mL of mannitol (1002 mOsm/L), at a dose of 2 mL/kg up to a maximum of 20 mL. Children were encouraged to void before solution administration, and a urine collection bag (Thermo Fisher Scientific, cat. #22275347, Rockford, IL) was placed and changed as needed for the following 5-hour period. Urine volume was measured following voiding and 1 two 2 drops of chlorhexidine (2.35%; Sigma-Aldrich, St Louis, MO) were added, and samples were aliquoted and stored on ice to limit bacterial growth. Urine aliquots were stored at −70°C before testing. Three additional 5-hour urine collections from children from the same cohort, who were between 3 and 15 months old and who had not been given the disaccharide solution on the same day, were also performed.

Between May 2011 and November 2012, a substudy was completed to determine the validity of L:M results between HPLC and LC-MSMS platforms. From among all of the L:M tests administered by May 2011, 100 samples, whose urinary lactulose and mannitol concentrations had previously been determined by the HPLC-PAD system described below, were selected for additional testing for lactulose and mannitol on 2 LC-MSMS systems. Testing in all 3 laboratories was blinded. Because it was hypothesized that accuracy may be associated with concentration, the sample selection included a subset of samples stratified by the concentration of analytes based on data from the HPLC laboratory. Because it was further hypothesized that interfering substances may be present that affect assay performance, and that the composition of urine may be age dependent, samples were also stratified by age. This included 5 relatively low and 5 relatively high-concentration urines from 3-, 6-, and 9-month-olds; and 9 relatively low- and high-concentration samples from 15-month-olds (39 total samples). In addition, 18 three-month urines, 19 six-month urines, 19 nine-month urines, and 5 fifteen-month urines were selected at random from a list of available specimens without regard for analyte concentration (61 randomly selected samples from a pool of ∼143–15 samples available per age point) (Fig. 1).

RESULTS

Recruitment into the cohort was ongoing between January 2010 and February 2012. Of the 303 infants initially enrolled, 49 contributed samples to the results reported here: 23 contributed one sample, 12 contributed 2 samples, 3 contributed 3 samples, and 11 contributed 4 samples. Descriptors of the overall cohort and the subset of infants whose samples were included in this comparison are shown in Table 1. All urine samples that had been collected and had HPLC results available by May 2011 were available for selection into the substudy (Fig. 1). Within the 100 samples selected, there were no indeterminate or missing results by any platform.

TABLE 1. Comparison of sociodemographic characteristics of cohort participants to the subset that contributed samples for analysis

Urinary Lactulose and Mannitol Concentrations

The log-transformed mean concentrations (millimoles per liter) of both mannitol and lactulose were statistically significantly different between LC-MSMS 1 and LC-MSMS 2, and between LC-MSMS 1 and HPLC (paired t tests). Overall for the 100 samples, the geometric mean mannitol concentrations were 1.56 mmol/L (95% CI 1.26–1.95) by LC-MSMS 1, 1.95 mmol/L (95% CI 1.54–2.48) by LC-MSMS 2, and 2.02 mmol/L (95% CI 1.60–2.56) by the HPLC-PAD platforms. Geometric mean lactulose concentrations were 0.58 mmol/L (95% CI 0.49–0.69), 0.64 mmol/L (95% CI 0.54–0.76), and 0.95 mmol/L (95% CI 0.80–1.12), respectively, and the geometric mean L:M ratios were 0.37 (95% CI 0.32–0.43), 0.33 (95% CI 0.28–0.39), and 0.46 (95% CI 0.49–0.56), respectively.

The HPLC-PAD system consistently overestimated mannitol concentration by approximately 34% (95% CI 26–41) at all ages. In contrast, lactulose concentration appeared overestimated to the greatest extent (by 207%, 95% CI 174–245) in the youngest infants. As a result, the L:M ratio was also overestimated to the greatest extent in the youngest children (Table 2).

TABLE 2. L:M ratios

The concentrations of the analytes in the 3 “matrix” urines (5-hour urine collections from children who were not given the sugar solution) ranged from 0.06 to 0.26 mmol/L mannitol and 0.11 to 0.46 mmol/L lactulose by LC-MSMS 1. Three of 61 of the randomly selected samples from dosed children had mannitol concentrations, as measured by LC-MSMS 1, in the range of the matrix urines, whereas 19/61 samples had lactulose concentrations in the range of the matrix urines.

Correlation Between Platforms

Spearman correlations were statistically significant between all 3 platforms for both analytes and the overall L:M ratio, both overall and for randomly selected samples (P < 0.05), in all instances. Between the HPLC-PAD and LC-MSMS 1 platforms for mannitol concentration, lactulose concentration, and the L:M ratio, these were 0.95, 0.74, and 0.75, respectively. Among randomly selected samples only (N = 61) this dropped to 0.95, 0.70, and 0.43, respectively (Fig. 2).

The Spearman correlations between the 2 LC-MSMS platforms were 0.97, 0.94, and 0.94 for mannitol concentration, lactulose concentration, and the L:M ratio, respectively; these dropped to 0.96, 0.89, and 0.89, when only randomly seeded samples were considered (Fig. 2).

Parametric correlations between the natural log of the mannitol concentration, lactulose concentration, and L:M ratio were also all statistically significant at the P < 0.05 level. For randomly selected samples, the correlations between the HPLC-PAD and LC-MSMS 1 platform results were 0.97, 0.63, and 0.33 for natural log of the mannitol concentration, lactulose concentration, and the L:M ratio, respectively, and 0.90, 0.96, and 0.81 between the 2 LC-MSMS platforms.

Overestimation at Low Lactulose Concentrations

When O/E ratios (HPLC-PAD result-LC-MSMS 1 result)/LC-MSMS 1 result were examined, samples with the greatest O/E ratios appeared to be clustered among samples with the lowest absolute lactulose concentrations as per LC-MSMS 1. The HPLC-PAD platform systematically overestimated concentrations at the low end of the spectrum, and this effect was present to a larger degree for lactulose (which was present at lower concentrations), than for mannitol. As a result, the L:M ratio was systematically inflated to the greatest extent among samples of the lowest true urinary lactulose concentrations (Fig. 3).

DISCUSSION

Within a sample of urine samples selected from 3- to 15-month-old infants in a cohort at risk of EE, overall agreement between the HPLC-PAD and LC-MSMS platforms for mannitol concentration was high (ρ ≥ 0.95), whereas agreement between the HPLC-PAD and LC-MSMS platforms for lactulose concentration and for the absolute L:M ratio was lower (ρ = 0.43–0.75).

Without communication between the laboratories and with different protocols, a commercial laboratory (LC-MSMS 2) was able to produce results that were highly correlated (ρ = 0.89) with the referent laboratory. Although the results could surely be brought closer with the central preparation and shipping of standards used for the assay and a standard protocol, it is notable that without these steps the precision of the platform was high.

Overall, the HPLC-PAD platform overestimated L:M ratios compared with the LC-MSMS 1 platform by 27%, whereas the commercial LC-MSMS platform underestimated them by 11%. In particular, in this instance, the HPLC-PAD platform tended to overestimate low concentration values compared with LC-MSMS, and this effect was more dramatic in lactulose than in mannitol. As a result of this effect, L:M ratios as measured by this platform tended to be higher among children who had had higher total urine volumes (who were also younger infants). When measured by LC-MSMS, there was no association between L:M and urine volume. This issue is relevant wherein the population of interest includes young infants with a developing capacity to concentrate urine, an age range of key interest in the field of EE.

Mannitol is known to be naturally present in urine in low concentrations, several times lower than the levels excreted in dosed children. In this instance, roughly 5% of randomly selected results had mannitol concentrations in the range of what was found in urine from undosed children. Lactulose is not found in natural foods; however, it is present in ultra-high-temperature milk in concentrations of 0.72 to 1.60 mmol/L 28. Children in the community from which these samples were drawn receive lactulose frequently in the form of yogurt and cereal preparations. Thirty-one percent of urinary lactulose concentrations were in the range of what was also found among undosed children (by LC-MSMS 1). Camilleri et al 23 have reported that excretion of lactulose continued between 8 and 24 hours after dosing in healthy adults; it is plausible that dietary lactulose was not fully cleared by the pre-urine collection fast. It also raises the possibility of interfering substances, most likely to be other dietary disaccharides or breakdown products of milk oligosaccharides most likely to be preferentially present in younger children who received breast milk during the challenge test.

Although the L:M test is the most commonly used assay for intestinal permeability and has been used for >2 decades, few standards have been adopted in implementation, the assessment of analyte concentrations, or the interpretation of results. Methodologically, dosages and urine collection times vary significantly between studies. For example, according to the protocols of 2 well-cited studies, the dosage of lactulose for a 10-kg infant would vary from 8 to 4 g and dosages of mannitol would be 2 and 1 g, respectively 2, 29, whereas in the present study, a child of the same size would receive 5 g lactulose and 2 g mannitol. Ideally, these probes would have no activity on the host in general and on the gastrointestinal system in particular; however, the doses of lactulose in these studies are in the pharmacologic range for its use as a hyperosmolar laxative and published protocols exceed the dose used for constipation and are in the range of that used for portal systemic encephalopathy 30. As a consequence, transit time is accelerated and this likely decreases the absorption of mannitol in the assay, whereas lactulose absorption is increased in the presence of hyperosmolar solutions 5, 31.

Because short (ie, 2-hour) urine collections are likely to reflect more small bowel absorption 23, this may be a more accurate test and more convenient for patient and provider. Logistically, however, 2-hour urine collections may be challenging, especially among children in the absence of catheter placement, because they may not void in this interval. The standardization and dissemination of protocols for dosage, including preferred saccharides, osmolarities, and the urine collection period best suited to EE research, are key next activities to advance the science in this field.

In summary, given the large analyte concentration range, the improved accuracy of LC-MSMS has important consequences for the assessment of lactulose and mannitol following oral administration in populations at risk for environmental or tropical enteropathy. Despite the increased cost of the instrument, the per-specimen cost is low and the platform is amenable to automation. We recommend that researchers wishing to implement a dual-sugar test as part of a study of EE use an LC-MSMS platform to optimize the accuracy of results and increase comparability between studies.