Overall, this meta-analysis demonstrates that despite considerable heterogeneity in the data, there are significant differences in LMR between healthy subjects and patients with either coeliac or Crohn’s disease. There are also significant differences in LMR between treated and untreated coeliac, and active compared to inactive Crohn’s disease.
These results hold true even when different L:M solute ratios are used. Altogether, there were 18 studies that used a 2:1 ratio of lactulose to mannitol, and 13 studies that used a 5:2 ratio. While no previous studies have performed direct comparisons of the data obtained using different solute ratios, we found that standard mean differences in LMR were larger when using the 2:1 ratio than the 5:2 (although high heterogeneity in the data means that this observation should be taken with caution, and statistical significance was not observed). However, the numbers in each subgroup were small, which may explain the mixed significances and heterogeneities observed. Nevertheless, this raises the possibility that the heterogenous nature of our results could be attributed to the different L:M solute ratios used, along with other factors such as assay method, time of fasting before the solute is administered, and urine collection times.
Musa et al. found no significant difference in LMR when comparing prolonged urine collection time (5 hours) with urine collected over a 2 hour period54. In the same study, they also found no significant differences when using two different analysis methods: high-performance anion exchange chromatography with derivatization-free, pulsed amperometric detection (HPAE-PAD); and liquid chromatography with tandem mass spectrometry (LC-MSMS)54. This concurred with results from Akram’s earlier study regarding urine collection, which found no significant differences in LMR when urine was collected over 2 and 6 hours55. Camilleri et al., on the other hand, found that LMR based on urine collections over 8–24 hours were significantly higher than those for collections times of 0–2 hours6. Interestingly, a study by Sequeira and colleagues suggested that differences in temporal patterns of excretion of lactulose and mannitol can be minimised if the urine collection period is restricted to 2½4 hours after solute ingestion56.
In relation to analysis platforms used for quantification of urinary lactulose and mannitol, Lee et al. found that LC-MSMS provides more accurate measurements than HPAE-PAD57. They subsequently recommended the former to be used in L:M studies. Nonetheless, current evidence surrounding the variable protocols used in L:M studies (e.g. in terms of the analysis platforms, solute ratios and urinary collection times used) is mixed, and more studies are required to elucidate the optimal method for performing this test.
Despite the variability and heterogeneity observed, we found significant differences in LMR between healthy controls and untreated coeliac disease. Patients with active coeliac disease are known to have flat mucosa, increased villous height, and increased paracellular permeability due to wider tissue junction pores and release of pro-inflammatory cytokines58, 59, 60. Gluten is thought to activate zonulin signalling, which opens up the tight junctions, causing increased paracellular permeability61. The role of gut permeability in the pathogenesis of coeliac disease is currently poorly understood, but it is thought that it might act to self-sustain the inflammatory response and perpetuate a vicious cycle62. Regardless, the increased permeability leads to an increase in lactulose excretion into urine, resulting in significantly higher LMR values than those observed in healthy subjects.
Differences in LMR between treated and untreated coeliac disease were also observed and found to be significant, with the change in LMR observed across all coeliac studies included in this review. These changes need to be analysed with caution, however, as the results were heterogenous and the 95% confidence interval was fairly wide (0.029–0.218).
There were also significant differences between the LMR values observed in treated coeliac disease and healthy controls, implying that it may take some time before mucosal integrity returns to baseline. A study by Cummins et al. showed (via the L:R test) that gut permeability improves after 2 months on a gluten free diet (GFD), but that it takes up to 6 months before villous recovery is observed63. Duerksen and colleagues demonstrated that more than 80% of coeliac patients on GFD for at least a year exhibited reduced gut permeability, although permeability only returned to normal levels in 48% of patients (10/21)64. Rajani et al. (one of the papers included in this analysis) reported no significant difference in LMR between healthy controls and coeliac patients who followed a GFD for a year44. Similarly, Vogelsang et al. (another paper included in this analysis) also reported no significant difference in LMR when comparing healthy subjects against coeliac patients who had a median of 44 months on a GFD35. However, LMR values were significantly different (healthy vs. treated coeliac) in the studies published by Vilela et al. (1 year of GFD)19 and Ukabam et al. (5–8 months of GFD)38. Thus, current evidence points towards the role of a GFD in improving gut permeability and restoring gut integrity after more than 12 months.
Another important finding in this review is the presence of significant differences in LMR between healthy controls and both active and inactive Crohn’s disease. Moreover, significant differences were also observed between active and inactive Crohn’s patients. While the pathogenesis of Crohn’s disease is multifactorial and the link to gut permeability is still not well understood, a study in 2019 using three-dimensional tissue culture models demonstrated that epithelial barrier dysfunction may be caused by Tumour Necrosis Factor (TNF)-α induced tight junction modulation and involvement of the c-Jun N-terminal protein kinase mitogen-activated protein kinases (JNK MAPK) signalling pathway65. Furthermore, altered gut permeability is surmised to be present at the early stages of disease, as increased paracellular permeability was found even in patients with quiescent IBD where endoscopic activity was absent66. Techniques other than the L:M test have also been used to assess gut permeability in Crohn’s disease. For example, in a recent study, a moderate positive correlation was found between excreted Chromium-52 labelled ethylenediamine tetraacetic acid (52Cr-EDTA) and faecal calprotectin levels (a known marker of gut permeability) in Crohn’s patients67. Similarly, the use of zonulin68 and Ussing chambers69 have also demonstrated increases in gut permeability in Crohn’s disease. Thus, our findings are in agreement with the above studies and provide further evidence for the importance of gut permeability in Crohn’s disease.
Interestingly, the differences in LMR between controls, active and inactive Crohn’s disease were smaller than those observed for untreated coeliac disease. This may indicate that the breakdown in epithelial barrier function is more pronounced in coeliac disease than it is in Crohn’s. Despite this, there was again significant heterogeneity in the difference values observed between control and active Crohn’s disease, and between inactive and active Crohn’s disease. This further indicates that observations made with the L:M test need to be taken with caution.
There are many advantages of the L:M test in measuring gut permeability. It is easy to perform, inexpensive and non-invasive. Our data also suggests that this test is associated with high sensitivity, making it a useful tool in screening for coeliac disease. (Sensitivity and specificity analysis was not performed for Crohn’s disease as the necessary data was not available in the papers included in our review). The L:M test was the method of choice in the MAL-ED study, which investigated the link between gut permeability and environmental enteropathy in children across 8 countries70. However, there is great variability in how the test is performed and our meta-analysis has revealed considerable heterogeneity in the results obtained. Interestingly, Ordiz et al. – who conducted the L:M test in 1669 rural Malawian children – surmised that the strong direct correlation between percentage lactulose and percentage mannitol excretion does not support the use of mannitol as a normalising factor for lactulose, and that using percentage lactulose excretion alone actually yields more information about gut integrity than LMR71. In addition, L:M measurements performed by Camilleri et al. indicated that LMR at 0–2 hours may in part reflect colonic permeability and not exclusively small bowel permeability. They have hence recommended that measurements of small bowel permeability using urine collected in the L:M test over 0–6 hours should be treated with caution72.
While there are clear limitations to the L:M test, the quantification of gut permeability in coeliac and Crohn’s disease reported here highlights a potential route for clinicians to better understand other gastrointestinal conditions where current diagnostics can be improved. For example, as larger changes in LMR were obtained in coeliac disease than in the Crohn’s disease (relative to healthy controls), this implies the possibility to stratify patients according to their gut permeability. Similarly, as differences were observed between treated and untreated patients, this suggests an opportunity to monitor for signs of relapse in a non-invasive manner (i.e. without the need for endoscopy). Thus, quantifying gut permeability may provide an avenue for the practising clinician to better assess patients with FGDs. Indeed, there is promise in utilising gut permeability values to improve management of this complex group of patients in either the primary care setting or the gastroenterology clinic. Nonetheless, we stress that the results of our meta-analysis do not necessarily suggest that the L:M test is currently suitable for this purpose. The high heterogeneity observed across groups and datasets means that it is unlikely that the L:M test will find widespread clinical use in its current form (and indeed explains why it has not done so to date). Hence, if assessment of gut permeability is to find widespread use in the diagnosis of FGDs, Crohn’s, coeliac or other conditions then it is highly likely that improved diagnostic tools/methods (or at the very least improved protocols for deployment of the L:M test) will be required.
The main limitation of this meta-analysis is the heterogeneity of our results, which is likely to be explained by the variations in how the L:M test was performed (and by physiological variations across individuals). A list of protocol variations that may have caused the heterogeneity is presented in Table S3. It is also difficult to directly assess the results of the two different solutes used due to the heterogeneity in most of these comparisons. In one of the studies18, both 5:2 and 2:1 lactulose: mannitol ratios in the solutes were given to patients during the L:M test, making it more difficult to compare the differences in LMR results between the two solutes. Furthermore, most studies that were available and included in the meta-analysis were at significant risk of bias. There are also variations in the way sensitivity and specificity were measured. As shown in Table S4, due to the heterogenous nature of the L:M test, different cut-off point values were used in included studies for diagnosis of disease. There were also not many studies that assessed the value of the L:M test as a screening measure in coeliac or Crohn’s disease.
Another limitation in this meta-analysis is in the variability in how active or inactive Crohn’s was defined as evidenced in Table S1. Crucially, however, most of these limitations are inherent to the L:M test itself. Thus, the fact that our results were heterogeneous highlights these important limitations to the L:M test and reveals that improvements are required if it is to be more widely used for clinical assessment of gut permeability.
Overall, this review has quantified the diagnostic value of the L:M test and reported the LMR values in healthy subjects, treated and untreated coeliac disease, and inactive and active Crohn’s disease. In addition, it provides a quantification of the heterogeneity in LMR values observed in these disease states. Our analysis demonstrates that there is potential value in measuring gut permeability in both Crohn’s and coeliac disease, and it provides an insight into the role of gut permeability in the pathogenesis of both conditions. Importantly, however, it also highlights the limitations in the L:M test and the need for both improved protocols and alternative diagnostic tools.