Sustained Diet-Induced Remission in Pediatric Crohn’s Disease Is Associated With Kynurenine and Serotonin Pathways

Abstract

Background

Both the Crohn’s disease exclusion diet combined with partial enteral nutrition (CDED+PEN) and exclusive enteral nutrition (EEN) can induce remission in mild-to-moderate pediatric Crohn’s disease and are associated with a marked decrease in fecal kynurenine levels. This suggests a link between clinical outcome of dietary therapy and changes in tryptophan metabolism pathways. Here, we characterize the changes in several fecal tryptophan metabolites induced by CDED+PEN or EEN and their association with remission.

Methods

A total of 21 tryptophan metabolites were quantified in fecal samples from a 12-week prospective randomized trial with CDED+PEN or EEN for induction of remission in mild to moderate pediatric Crohn’s disease. Tryptophan metabolites at week 0 (W0), W6, and W12 of 73 samples were quantitatively measured by liquid chromatography coupled with triple quadrupole mass spectrometry, and data were analyzed according to clinical groups of baselines (W0), induced remission at W6, no remission, sustained remission at W12, and nonsustained remission.

Results

Reduction in components of the kynurenine pathway, such as kynurenine and quinolinic acid, were strongly associated with induced remission with both CDED+PEN and EEN, which were maintained in sustained remission. Specific serotonin pathway metabolites, such as melatonin, N-acetylserotonin, and 5-OH-tryptophan, were significantly increased in fecal samples from patients maintaining remission at W12 with both CDED+PEN and EEN. Importantly, in samples from patients failing to sustain remission, no changes were observed. Remission induction with EEN differs from CDED+PEN, particularly the moderate effects on indole pathway metabolites. The ratios of kynurenine and melatonin and quinolinic acid and melatonin perform well as markers for sustained remission.

Conclusions

The reduction in specific kynurenine pathway compounds and the increase in serotonin pathway compounds are associated with diet-induced and sustained remission. Further studies are warranted to assess causality and the association of these metabolites with specific diet and lifestyle factors, affecting sustained clinical remission.

Key Messages.

What is already known?

• The balance between the tryptophan metabolism pathways (kynurenine and serotonin pathways) is critical to maintaining intestinal homeostasis. Elevated concentrations of kynurenine pathway metabolites and the reduction in some serotonin pathway metabolites are strongly associated with intestinal inflammation.

What is new here?

• This study suggests that the degradation of tryptophan into the kynurenine metabolism pathway is reduced in favor of its degradation into the serotonin metabolism pathway as inflammation improves with dietary therapy in pediatric Crohn’s disease.

How can this study help patient care?

• These data may inform future research to improve dietary intervention in pediatric Crohn’s disease and could help identify new therapeutic targets and metabolic biomarkers.

Introduction

The increased incidence of pediatric Crohn’s disease (CD) worldwide has been strongly linked to dietary shifts toward a Westernized diet,^1-3 which includes high amounts of processed foods, red meat, high fat, sugar and additive exposure, and a lack of dietary fiber, fruit, and vegetables.³ Westernized diet patterns may negatively affect the gut microbiome and metabolome and alter the intestinal immunity, which influences CD development and the disease course.³ Sufficient tryptophan intake is necessary to maintain gut homeostasis: a diet deficient in tryptophan has been shown to alter the gut microbiota and increase inflammation.⁴ The CD exclusion diet combined with partial enteral nutrition (CDED+PEN) and exclusive enteral nutrition (EEN) were designed to minimize dietary exposure to foods that might adversely influence the microbiome, metabolome, intestinal barrier, and immunity.¹

Although dietary therapies are effective in inducing remission and have been recommended as first-line therapy for mild-to-moderate pediatric CD, there are still some challenges.^1,5,6 For EEN therapy, rigorous exclusion of solid food intake is necessary for success, which makes EEN an unsuitable treatment to maintain remission long term.^1,5-7 In addition, and as repeatedly shown, there is a rapid return of inflammation upon food reintroduction.^1,8-10 CDED+PEN is primarily based on table food with improved tolerance compared with EEN and has also shown promise to sustain remission. Like any behavioral modification, CDED+PEN requires a strong commitment to planning, preparation, and cost of meals according to the dietary instructions.⁵ Thus, understanding the mechanism of actions of nutritional therapy, including specific changes in metabolomics or microbiome associated with clinical remission induction and maintenance, may yield new strategies to simplify and ultimately improve the dietary intervention. This could potentially identify new therapeutic targets and perhaps even prevention approaches.

In humans, the main source of tryptophan is dietary intake. Tryptophan is metabolized by 3 pathways in the gastrointestinal tract: kynurenine, serotonin, and microbiota-derived indole pathway metabolites.^11,12 The balance between these 3 pathways is critical to maintaining intestinal homeostasis.¹¹ Alterations in tryptophan metabolism pathways are strongly associated with CD, suggesting an active role in disease development.^2,11,13-16 Elevated concentrations of some kynurenine pathway metabolites, such as kynurenine and quinolinic acid, have been detected in fecal and serum samples of inflammatory bowel disease (IBD) patients.^2,13 Moreover, increased serotonin levels and a reduction in some serotonin pathway components such as melatonin and 5-OH-tryptophan were seen in colitis.^14-16

We have recently shown significant changes in a limited set of fecal metabolite levels that were associated with clinical outcome and type of dietary intervention.² Among the most robust associations with remission was the reduction in kynurenine levels with both CDED+PEN and EEN. The changes in kynurenine levels displayed the highest discriminatory biomarker power compared with other metabolite changes.² While these data suggest a potential link between possible changes in tryptophan metabolism pathways and the clinical outcome associated with the dietary therapy, there were a limited number of tryptophan metabolites included in our previous study. In addition, the changes were mainly displayed as normalized intensity and not absolute concentrations, which limited the interpretation of the data.

In the present study, we sought to characterize the changes in the absolute concentration of multiple metabolites belonging to the 3 different tryptophan metabolism pathways (the measured metabolites are illustrated in Figure 1) in the feces of pediatric CD patients that were subjected to CDED+PEN or EEN therapy. We aimed to analyze the data according to the clinical outcome and type of dietary therapy and identify potential biomarkers associated with clinical remission.

Figure 1.

The chemical structure and the position of all measured metabolites in different tryptophan metabolism pathways baseline (W0), induced remission at W6 (W6_rem), no remission (W6_nr), sustained remission at W12 (W12_sr), and nonsustained remission (W12_nsr).

Methods

Study Design

A total of 21 tryptophan metabolites were quantified in 73 fecal samples collected from 43 pediatric CD patients through a previously performed 12-week prospective randomized clinical trial (RCT)^1,2 using 2 different nutritional therapies (CDED+PEN: n = 22 patients; or EEN: n = 21 patients) for induction of remission in mild-to-moderate pediatric CD.

The first group followed CDED, with additional PEN (MODULEN IBD; Nestlé) for 50% of calculated energy requirements for 6 weeks, followed by a step-down diet with an additional 25% of energy requirements by PEN until week 12 as described previously.^1,2 The second group followed 100% EN for 6 weeks, after which they gradually returned to a free diet with 25% of calories from PEN as described previously.^1,2 The primary outcome of the RCT was the tolerability to the dietary therapy. Secondary endpoints of the study were intention-to-treat remission at week 6 (defined by a Pediatric Crohn’s Disease Activity Index score ≤10) and corticosteroid-free intention-to-treat sustained remission at week 12. The study was approved by the local ethics board at each of the participating sites, and informed consent was obtained for each participant as described previously (NCT01728870).^1,2

Subjects

Each patient was provided 2 containers labeled with stickers. When the patient brought the samples (up to 24 hours), we kept the calprotectin container at -20 °C and split the microbiome/metabolome container into 4 transfer tubes (Eppendorf tube with black cap) using a wooden stick and stored them immediately at -80°C without any buffer.

A targeted quantitative measurement of tryptophan metabolites at week 0 (W0), W6, and W12 of 73 fecal samples was performed by liquid chromatography coupled with a quadrupole mass spectrometry as described previously.¹⁷ The data were analyzed according to clinical groups of baseline (W0), induced remission at W6 (W6_rem), no remission (W6_nr), sustained remission at W12 (W12_sr), and nonsustained remission (W12_nsr).

The number of CDED+PEN fecal samples was the following (total: n = 38): W0 (n = 14), W6_rem (n = 13), W6_nr (n = 2), W12_sr (n = 8), and W12_nr (n = 1). The number of EEN fecal samples was the following (total: n = 35): W0 (n = 13), W6_rem (n = 9), W6_nr (n = 4), W12_sr (n = 6) and W12_nr (n = 3). The baseline characteristics of patients included in this RCT are described in Table 1 (updated from Ghiboub et al).²

Table 1.

Baseline characteristics of patients included in the clinical study.

Characteristic/Variable	CDED+PEN (n = 22)	EEN (n = 21)	P Value
Age, y	14.3 (2.8)	14.3 (2.5)	.949
Male	14 (64)	14 (67)	1
Disease duration, d	31 (11-397)	27 (14-76)	.504
PCDAI	22.5 (17.5-33.1)	27.5 (16.3-32.5)	.788
CRP, mg/L	16.4 (6.6-41.6)	21.1 (9.7-39.6)	.659
Immunomodulators	3	4	.689
Disease location (Paris classification)			.831
L1	9	7	—
L2	1	0	—
L3	11	12	—
L4a	9	9	—
L4b	2	2	—
L4a + b	3	3	—
Good compliance	18	16	.721
Patients with available fecal sample at each time point per diet
W0	14	13
W6	15	13
W12	9	9
Patients with available fecal sample in remission at W6 or sustained remission at W12 per diet group (based on PCDAI score <10)
W6	13/15 (87)	9/13 (69)
W12	8/9 (98)	6/9 (66)

Values are mean ± SD, n (%), median (interquartile range), n, or n/n (%).

Abbreviations: CDED+PEN, Crohn’s disease exclusion diet combined with partial enteral nutrition; CRP, C-reactive protein; EEN, exclusive enteral nutrition; PCDAI, Pediatric Crohn’s Disease Activity Index; W, week.

Fecal Tryptophan Metabolites Quantification

For fecal tryptophan metabolites measurement, 3 mg of lyophilized stool samples and 900 μL of MeOH/H₂O mixture (1:1) were added according to internal standards. The samples were vortexed, homogenized, and centrifuged. Then, 100 µL of the supernatants were transferred to a 96-well plate for metabolites quantification and 800 µL were evaporated to dryness with a vacuum concentrator (Savant SPD 111 vs SpeedVac; Thermo Fisher Scientific) at 45°C for 4 hours. The residues were reconstituted in 100 μL of MeOH/H₂O mixture (1:9), and 2 µL were analyzed.

The liquid chromatography tandem mass spectrometry quantification was performed as previously described.¹⁷ A calibration curve was created for each metabolite by calculating the intensity ratio obtained between the metabolite and its internal standard. These calibration curves were then utilized to determine the concentrations of each metabolite in fecal samples. Specific calibration ranges and internal standards concentrations associated to fragmentation parameters are provided in the Supplemental Table 1.

The 21 tryptophan metabolites measured were tryptophan, kynurenine pathway metabolites (kynurenine, kynurenic acid, anthranilic acid, 3-OH-kynurenine, 3-OH, anthranilic acid, picolinic acid, and quinolinic acid), serotonin pathway metabolites (5-OH-tryptophan, serotonin, N-acetylserotonin, melatonin, and 5-OH-indoleacetic acid), and indole pathway metabolites (tryptamine, indoxyl sulfate [indole-3-sulfate], indole-3-acetamide, indole-3-acetic acid, indole-3-lactic acid, indole-3-aldehyde, tryptophol, and indole).

The chemical structure and the position of all measured metabolites in different tryptophan metabolism pathways are illustrated in Figure 1.

The list of tryptophan metabolites was selected based on their reported disturbances of their concentrations in multiple inflammatory diseases including IBD.^11,12

Data Preparation and Statistical Analysis

Data illustration and statistical analysis were performed with MetaboAnalyst5.0 (https://www.metaboanalyst.ca).¹⁸ The data were filtered (using the interquartile range method) to identify and remove noisy and noninformative variables that are unlikely to be of use when modeling the data as described prevously.¹⁹ Receiver-operating characteristic (ROC) analysis was also performed in MetaboAnalyst 5.0¹⁸ as described previously²⁰ using calculated normalized concentrations of 20 tryptophan metabolites. The area under the ROC curve (AUC) was calculated with 95% confidence intervals. Ratios between 2 tryptophan metabolite concentrations may yield more information than the 2 corresponding metabolite concentrations alone. Thus, we commanded MetaboAnalyst 5.0 to compute and include all possible ratios of the top 20 tryptophan metabolite pairs (based on P values) to be integrated in further biomarker analysis. Individual metabolites and metabolite ratios with AUC ≥0.8 and P < .05 were selected. The final figures were adapted in Inkscape 0.92.4.

For group analysis, data were subjected to 1-way analysis of variance or Student’s t test. The 2-tailed level of significance was set at P ≤ .05. Data are shown as mean ± SEM.

Results

Baseline Characteristics of Patients and Rate of Remission per Group

There were no significant differences in baseline characteristics of patients at W0, Pediatric Crohn’s Disease Activity Index score (Table 1) or tryptophan metabolite concentrations between CDED+PEN and EEN samples at baseline (W0), indicating a well-randomized study. By including only the patients with available fecal samples at specific time points, CDED+PEN induced remission in 13 (87%) out of 15 patients at W6, which was maintained in 8 (89%) out of 9 patients at W12 (Table 1). In the EEN group, among 13 patients, 9 (69%) achieved remission at W6, and 6 (67%) out of 9 patients maintained remission until W12 (Table 1).

Tryptophan Metabolite Changes with CDED+PEN–Induced and CDED+PEN–Sustained Remission

We analyzed the association between CDED+PEN–induced and CDED+PEN–sustained remission and tryptophan metabolites composition. Of the 21 tryptophan metabolites included in the analysis, 2 significant changes (P ≤ .05) were associated with W6_rem (Figure 2A), and 4 were observed with W12_sr (Figure 3B). The reduction of components of the kynurenine pathway; kynurenine (P = 9.67 × 10^-4) (Figure 2A and 2C) and quinolinic acid (P = .04) (Figure 2A and 2C) were strongly associated with W6_rem. This reduction was maintained through W12_sr (P = .003 and .02 for kynurenine and quinolinic acid, respectively) (Figure 2B and 2C).

Figure 2.

Crohn’s disease exclusion diet combined with partial enteral nutrition (CDED+PEN)–induced and CDED+PEN–sustained remission were associated with specific changes in some kynurenine and serotonin pathway metabolites. Volcano plot illustrating the changes in tryptophan metabolites comparing (A) induced remission at week 6 (W6_rem) vs W0 or (B) sustained remission at W12 (W12_sr) vs W0. Blue dots represent significantly decreased metabolites at W6_rem or W12_sr compared with W0, based on t test (≤0.05). Red dots represent the significantly increased metabolites. Black dots represent no significant changes. (C) The absolute concentrations (pmol/mg) of the significantly changed fecal tryptophan metabolites at W0, W6_rem, and W12_sr (≤0.05). W0 (n = 27), W6_rem (n = 13), and W12_sr (n = 8).

Figure 3.

Exclusive enteral nutrition (EEN)-induced and EEN-sustained remission were associated with specific changes in some kynurenine, indole, and serotonin pathway metabolites. Volcano plot illustrating the changes in tryptophan metabolites comparing (A) induced remission at week 6 (W6_rem) vs W0 or (B) sustained remission at W12 (W12_sr) vs W0. Blue dots represent significantly decreased metabolites at W6_rem or W12_sr compared with W0, based on t test (≤0.05). Red dots represent the significantly increased metabolites. Black dots represent no significant changes. (C) The absolute concentrations (pmol/mg) of the significantly changed fecal tryptophan metabolites at W0, W6_rem, and W12_sr (≤0.05). W0 (n = 27), W6_rem (n = 9) and W12_sr (n = 6).

The concentration of some serotonin pathway metabolites such as melatonin (P = .001) and N-acetylserotonin (P = .01) were significantly and selectively increased with W12_sr (Figure 2B and 2C). No significant changes in other tryptophan metabolites were observed in both CDED+PEN–induced and CDED+PEN–sustained remission fecal samples.

Tryptophan Metabolite Changes with EEN-Induced and EEN-Sustained Remission

We then analyzed the impact of EEN-induced and EEN-sustained remission on tryptophan metabolites composition. Of the 21 tryptophan metabolites included in the analysis, 5 significant changes (P ≤ .05) were associated with W6_rem (Figure 3A), and 4 associated with W12_sr were observed (Figure 3B).

In line with CDED+PEN, EEN fecal samples also demonstrated a significant decrease in kynurenine (P = .04) and quinolinic acid (P = .03) levels with W6_rem (Figure 3A), which was sustained with W12_sr (Figure 2B and 2C). The drop in fecal serotonin level and the increase in indole level characterized only W6_rem (P = .01 and .03, respectively) (Figure 3A) and not W12_sr, which were comparable to baseline level.

Similar to observations in CDED+PEN W12_sr, the increase in the serotonin pathway compounds, melatonin and N-acetylserotonin (normelatonin) (P = .04 and .054, respectively), were also associated with the clinical sustained remission at W12 (Figure 3B and 3C). Another serotonin metabolite, 5-OH-tryptophan, was significantly increased with EEN W12_sr (P = .02) (Figure 3B and 3C). There were no significant changes in other tryptophan metabolites.

Tryptophan Metabolite Profile in EEN Nonresponders or Early Flare After EEN

Around 30% (n = 4 of 13) of patients on EEN did not achieve remission at W6 and 33% (n = 6 of 9) did not sustain remission (Table 1). This prompted us to explore whether the tryptophan metabolite profile in these nonresponding or early flare samples is comparable to baseline. With the exception of quinolinic acid (which was still significantly reduced), no significant changes were detected in serotonin, kynurenine, indole, or melatonin levels with W6_nr samples compared with W0 (Supplementary Figure 1) (in contrast with W6_rem samples).

In samples from patients failing to sustain remission (EEN W12_nsr), no significant changes were observed compared with W0 including kynurenine, quinolinic acid, and serotonin pathway metabolites (Figure 4A). This observation was confirmed when we analyzed single tryptophan metabolites across different conditions: W0 vs W12_sr vs W12_nsr (Figure 4B). The changes in some metabolites that were highly associated with W12_sr such as melatonin and 5-OH-tryptophan were comparable to baseline levels with W12_nsr samples (Figure 4B).

Figure 4.

Nonresponder and early flare samples (with exclusive enteral nutrition [EEN]) display similar tryptophan metabolite profile to baseline samples. (A) Volcano plot illustrating the changes in tryptophan metabolites comparing nonsustained remission (W12_nsr) vs week 0 (W0). (B) The absolute concentrations (pmol/mg) of melatonin and 5-OH-tryptophan at W0, no remission (W6_nr), and W12_nsr. Black dots represent no significant changes. W0 (n = 27), induced remission at W6 (n = 9), sustained remission at W12 (W12_sr) (n = 6), and W12_nr (n = 3). FC, fold change.

Tryptophan Metabolite Ratios as a Potential Biomarker Associated With Clinical Outcome

ROC analysis computed for the 20 tryptophan metabolites showed strong signatures associated with CDED+PEN–induced, CDED+PEN–sustained, EEN-induced, or/and EEN-sustained remission (Supplemental Tables 2-5). Most of the AUCs above 0.8 were obtained for ratios between the concentration of 2 metabolites: only a few individual metabolites showed AUC > 0.8 (Supplemental Tables 2-5). For instance, by applying the cutoff of AUC >0.8 and P ≤ .05; 8, 16, 5, and 15 biomarker signatures were detected in CDED+PEN W6_rem, CDED+PEN W12_sr, EEN W6_rem, and EEN W12_sr, respectively (Supplemental Tables 2-5).

In general, ratios between certain kynurenine pathways metabolites and melatonin were the most promising signatures associated with CDED+PEN–induced and/or CDED+PEN–sustained remission. The kynurenine-to-melatonin ratio (AUC, 0.886) (Figure 5A andSupplemental Table 2) performed better than kynurenine alone (AUC, 0.818) (Figure 5B andSupplemental Table 2) as a signature for CDED+PEN W6_rem. The kynurenine-to-tryptophan ratio, which has been previously reported to be increased in CD patients,² was significantly decreased with induced remission in CDED+PEN with an AUC of 0.8 (Supplemental Table 2). The kynurenine-to-melatonin (AUC, 0.968) (Figure 5C) and quinolinic acid-to-melatonin (0.944) ratios (Figure 5D andSupplemental Table 3) performed remarkably better as biomarker signatures with CDED+PEN W12_sr. The individual metabolite melatonin performed well too but less performant compared with ratios (Supplemental Table 3).

Figure 5.

Biomarker signatures associated with Crohn’s disease exclusion diet combined with partial enteral nutrition (CDED+PEN)–induced and CDED+PEN–sustained remission. The receiver-operating characteristic (ROC) curves with their area under the ROC curve (AUC) scores and the associated box plot of raw values with their P value of (A) the kynurenine-to-melatonin ratio and (B) the single kynurenine distinguishing CDED+PEN induced remission at week 6 (W6_rem) from W0 baseline. ROC curves with their AUC scores and the associated box plot of raw values with their P value of (C) the kynurenine-to-melatonin ratio and (D) the quinolinic acid-to-melatonin ratio distinguishing CDED+PEN sustained remission at W12 (W12_sr) from W0 baseline. W0 (n = 27), W6_rem (n = 13), and W12_sr (n = 8). Box plot represents the distribution of the kynurenine or metabolite ratios used for model building.

The ratios between serotonin and some other serotonin pathway components such as N-acetylserotonin (AUC, 0.893) (Figure 6A andSupplemental Table 4) or melatonin (AUC, 0.848) (Figure 6B andSupplemental Table 4) displayed the highest biomarker signature power associated with EEN W6_rem. The kynurenine-to-melatonin (AUC, 0.914) (Figure 6C andSupplemental Table 5) and quinolinic acid-to-melatonin (AUC, 0.981) (Figure 6D andSupplemental Table 5) ratios performed very well as biomarker signatures for sustained clinical remission (W12_sr) with EEN. Other promising biomarker signatures are listed in Supplemental Tables 2-5.

Figure 6.

Biomarker signatures associated with exclusive enteral nutrition (EEN)-induced and EEN-sustained remission. The receiver-operating characteristic (ROC) curves with their area under the ROC curve (AUC) scores and the associated box plot of raw values with their P value of (A) the ratio of serotonin and N-acetylserotonin and (B) the ratio serotonin and melatonin distinguishing EEN W6_rem from W0 baseline. ROC curves with their AUC scores and the associated box plot of raw values with their P-value of (C) the ratio quinolinic acid and melatonin and (D) the ratio kynurenine and melatonin distinguishing CDED+PEN sustained remission at W12 (W12_sr) from W0 baseline. W0 (n = 27), induced remission at W6 (W6_rem) (n = 9), and W12_sr (n = 6). Box plot represents the distribution of the kynurenine or metabolite ratios used for model building.

Discussion

We have shown that clinical outcomes of dietary therapy are strongly linked to specific changes in some kynurenine and serotonin pathway compounds that were previously shown to be associated with mediating intestinal homeostasis and immunity, as well as mediating mood and sleep regulation.^2,13-15,21 In this study, we report the reduction in kynurenine pathway metabolites, which appear early with clinical remission and are maintained with ongoing dietary therapy and clinical sustained remission, while the increase in some specific serotonin pathway compounds occurs with the clinical outcome of sustained remission.

Importantly, these observations were not seen when an early flare occurred upon resuming a free diet in the EEN group. These data suggest that dietary therapy–induced and sustained remission may be linked with certain tryptophan metabolite composition level changes that dietary therapy nonresponders or patients experiencing an early flare upon resuming free diet do not display.

Induced and sustained remission with EEN differs from CDED+PEN with changes in some indole pathway metabolites. Ratios between kynurenine or quinolinic acids and melatonin performed remarkably well as biomarkers for sustained remission with both dietary interventions. This suggests that some metabolic changes might reflect remission, whereas others could represent diet-specific processes.

Changes in Kynurenine Pathway Metabolites

The main source of kynurenine is from the degradation of tryptophan. Kynurenine is further metabolized by several host enzymes, eventually leading to the formation of quinolinic acid.^11,12 The tryptophan metabolites (kynurenine and quinolinic acid) have been suggested to be critical in modulating intestinal immunity and disturbances in their fecal and serum concentrations were linked to CD.^2,13 Several studies of tryptophan metabolites reported an overproduction of kynurenine and quinolinic acid in inflammatory diseases. Based on previous literature, it is suggested that during inflammation these metabolites are increased in order to suppress inflammation rather than being proinflammatory metabolites themselves.^2,13 For instance, kynurenine has been suggested to play a critical role in controlling the inflammatory response via activating aryl hydrocarbon receptor (AhR) signaling.¹² AhR activation has been shown to have an anti-inflammatory effect. Quinolinic acid has been suggested to have immunosuppressant properties (eg, on cytokine production).²²

In this study, both CDED+PEN and EEN therapies were strongly associated with reduced kynurenine and quinolinic acid compared with baseline samples. This reduction persists through W12_sr. Kynurenine reduction in this study confirmed our previous observations.² Thus, CDED+PEN–induced, CDED+PEN–sustained, EEN-induced, and EEN-sustained remission point to either kynurenine and quinolinic acid reductions as one of the factors that may functionally participate in the clinical outcome of remission, or just reflecting the reduction in inflammatory response due to diet-induced remission. In active IBD, IDO1 is highly expressed in cells of both the lamina propria and epithelium.²³ IDO1 is critical to trigger the kynurenine pathway, and its decrease with reduced inflammation can affect kynurenine pathway metabolites synthesis in colitis.^24,25

Changes in Serotonin Pathway Metabolites

Changes in some serotonin pathway metabolites have been consistently observed in studies of experimental colitis and IBD patients, particularly the reduction in 5-OH-tryptophan, N-acetylserotonin, and melatonin.^14-16 These metabolites were significantly increased with dietary therapy–sustained remission in our study but not in samples in which remission was not sustained, suggesting a role in mediating intestinal homeostasis.

While no data are yet available in colitis, administration of 5-OH-tryptophan led a strong anti-inflammatory effect in collagen-induced arthritis, through decreasing the production of proinflammatory cytokines and reducing T helper 1 and T helper 17 populations.²¹ Interestingly, this was associated with decreased serum levels of kynurenine,²¹ which corroborates our observation in feces with dietary therapy–induced remission in CD.

Melatonin supplementation has been found beneficial in some preclinical studies of IBD but has been understudied in clinical IBD.^14,26,27 Melatonin and its direct precursor N-acetylserotonin have demonstrated strong antioxidant properties and anti-inflammatory effects.²⁸ The effect of melatonin on shaping gut microbiota composition in relation to susceptibility to colitis is well established.¹⁴ Kim et al¹⁴ demonstrated that melatonin reverts microbial dysbiosis by increasing the ratio of Firmicutes to Bacteriodetes through controlling antimicrobial peptide production in DSS-induced colitis.

We previously showed that CDED+PEN– and EEN-induced remission is associated with improvement of dysbiosis, but some dysbiotic features remain even after 12 weeks of treatment.^1,7 In our current study, we show that dietary therapy–sustained remission is associated with markedly increased levels of melatonin and N-acetylserotonin. Thus, we suggest the changes in melatonin and its precursor N-acetylserotonin level might be a potential player in the remission maintenance by positively affecting the gut microbiota composition and controlling the inflammatory response. It is tempting to speculate that these metabolites also play a role in the improved quality of life associated with clinical remission.^29,30

Serotonin is produced mainly in the gastrointestinal tract,³¹ and its reduction in untreated CD patients has been suggested to influence CD development.³² Serotonin influences autophagy and gut microbiota composition and is associated with disease activity.^33,34 We found significant associations in this study, with increased N-acetylserotonin (also known as normelatonin, a precursor of melatonin in the production pathway from serotonin) in CDED+PEN–sustained remission, whereas a significant decrease in fecal serotonin level was only associated with EEN-induced remission at week 6. These observations suggest that pairwise investigations of metabolites of these metabolic pathways may shed more light on the true strength and biological relevance of these associations.

Biomarker Signatures

ROC analysis revealed robust biomarker signatures associated with the clinical outcome of remission with both dietary therapies. In particular, ratios between kynurenine pathway metabolites (kynurenine or quinolinic acid) and the serotonin pathway metabolite (melatonin) performed well as biomarkers for remission with CDED+PEN and maintaining remission with both CDED+PEN and EEN therapy. Interestingly, the ratios between serotonin and some other metabolites in the serotonin pathway such as N-acetylserotonin and melatonin were the most promising signatures associated with EEN-induced remission samples. These data can add to the development of more comprehensive disease signatures combining calprotectin, microbiome, and metabolomics to define and monitor clinical outcomes with dietary therapy.

While the ROC analysis is meant to identify the accuracy of biomarkers for dietary therapy–induced remission, the patterns of different tryptophan pathways activation we identified may give some indications about the mechanism of action of nutritional therapy. The serotonin-to-N-acetylserotonin and serotonin-to-melatonin ratios were high at baseline and significantly decreased with EEN-induced remission. Serotonin is the main precursor for N-acetylserotonin and melatonin synthesis, which indicates that the degradation of serotonin into subsequent metabolites in the serotonin metabolism pathway is enhanced with remission induction.

Ratios between kynurenine and melatonin and quinolinic acid and melatonin were significantly reduced with sustained remission compared with baseline, suggesting that the degradation of tryptophan into the kynurenine metabolism pathway is reduced in favor of its degradation into the serotonin metabolism pathway as inflammation improves. In CD patients, previous studies of active disease have shown an accumulation of serotonin^33,34 and reduction in serotonin pathway metabolism activation^11,14-16 and an increased kynurenine pathway activation leading to kynurenine and quinolinic acid accumulation.^2,11,13 Thus, our data suggests that dietary therapy may correct the inflammatory disturbance in the different tryptophan pathways.

As the kynurenine and serotonin pathways generally are not directly related to intestinal microbiota, this suggests that nutritional therapy may indeed affect some host enzymes regulating these pathways such as IDO1 and TpH1.^11,12 In turn, this can be the consequence of the reduced inflammation by nutritional therapy. Transcriptional studies and the assessment of tryptophan-metabolizing enzymes are warranted to better understand the effects of diet-induced remission on tryptophan metabolism. For instance, Wnorowski et al³⁵ showed that in intestinal tissues of active IBD, the changes in the expression of genes encoding the tryptophan-metabolizing enzymes indicate the shift in intestinal tryptophan metabolism from the synthesis of serotonin pathway metabolites toward kynurenine pathway metabolites. Importantly, these transcriptomic profiles were normalized after successful treatment with infliximab (responders),³⁵ which corroborates our finding using dietary therapy.

The intestinal microbiota orchestrates the direct tryptophan degradation into indole pathway metabolites and derivatives.^11,12 While the beneficial clinical outcome of both CDED+PEN and EEN was associated with strong changes in the concentration of some key compounds of kynurenine and serotonin pathways, the concentrations of indole pathway metabolites were unchanged in CDED+PEN samples, and only moderate changes in indole pathway metabolites were observed with EEN. This corroborates our previous microbiome (metagenome) findings on the same fecal samples that were used for this targeted tryptophan metabolites study.² No significant differential gene abundances were detected in indole pathway comparing diet-induced remission vs nonremission.²

In this study, we used fecal samples from a previously reported RCT^1,2 to describe the impact of CDED+PEN and EEN on gut tryptophan metabolite profiles and the association with the clinical outcomes. Limitations of our study include the small sample size, particularly concerning CD patients not achieving or maintaining remission with CDED+PEN, as around 88% achieved and sustained remission with CDED+PEN from W6 to W12. We used a targeted approach to measure 21 tryptophan metabolites complementing previously presented targeted investigations of metabolomics composition and microbiome.^1,2 As tryptophan pathways contain many other metabolites (not included in this study), this inevitably leads to reduced resolution of the analysis of complex metabolic pathways.

Conclusions

This study shows that a reduction in kynurenine pathway activity and an increase in serotonin pathway activity are associated with diet-induced and -sustained remission. These data may inform future research to improve dietary interventions in CD. The pairwise metabolite investigations we presented show promise as markers of metabolomic diet–induced remission and warrant further study in conjunction with other noninvasive inflammatory and microbiome markers. Future study to understand the effect of the identified tryptophan metabolite changes on achieving remission through anti-inflammatory and regulatory effects, eg, through AhR signaling and tryptophan metabolism–associated genes and enzymes, is warranted.

Author Contribution

Study design: J.E.V.L. Laboratory assay: B.S., M.G. Analysis and data preparation: MG. Writing of the manuscript: M.G., R.S.B. Metabolites quantification: A.L., P.E. Figures and Tables preparation: M.G., C.M.V. Supervision: J.E.V.L. and W.J.D.J. Reviewing the manuscript: J.E.V.L., R.S.B., E.W., C.M.V., W.J.D.J., B.S., and M.A.B. All authors contributed to the article and approved the submitted version.

Funding

The conduct of the study in Canada (Halifax, Edmonton) was supported by local divisional funds, a Women and Children’s Health Research Institute, Research Capacity Building Award, a Canadian Institutes of Health Research New Investigator award, and a Canada Research Chair Tier 2 in Translational Microbiomics (to J.E.V.L.). J.E.V.L. was supported by funding from the Amsterdam UMC; Wetenschappelijke Advies Raad of Stichting Steun Emma (Emma Children’s Hospital); the Department of Pediatrics, Amsterdam University Medical Centers; and by a Pro-KIIDS Clinical Research Network Award (#585718).

The funders of the study had no role in the design of the study, data collection or analysis, interpretation of data, writing of the report, or decision to submit the article for publication. None of the funders had access to the data.

Conflicts of Interest

The authors declare no potential conflict of interest.