Prebiotic fiber enteral supplementation after allogeneic transplantation: feasibility and impact on the microbiome Open Access

During allogeneic hematopoietic stem cell transplantation (HSCT) there are many factors that cause gastrointestinal microbiome dysbiosis, including pre-HSCT conditioning and antibiotics given as prophylaxis or for febrile neutropenia.^1,2 Post-HSCT microbial diversity decreases,^3,4 with an unfavorable change in dominating taxa^2,3 and reduction in short chain fatty acid (SCFA) levels.⁵ Lower microbial diversity before HSCT is associated with higher risk of pulmonary complications⁶ and bloodstream infections,⁷ whereas lower diversity after transplant has been associated with increased graft-versus-host disease (GVHD)^5,8 and mortality.^9,10 The gut microbiome can also act as a reservoir for antibiotic resistance, and the gut microbiota resistome has been associated with febrile neutropenia.¹¹ 

Dietary fiber, particularly prebiotic fiber, directly influences the microbiome. Prebiotics, including fibers such as inulin, are substrates that are selectively used by microorganisms conferring health benefits¹² and a high-fiber diet is associated with higher microbial diversity.¹³ SCFAs such as butyrate, propionate, and acetate produced through bacterial fermentation of prebiotic fiber support gastrointestinal barrier maintenance and immune function, have an anti-inflammatory effect, and suppress the growth of pathogens.¹⁴ In patients undergoing HSCT, higher abundance of butyrate producing bacteria has been associated with fewer respiratory tract infections,¹⁵ and low SCFA levels linked with GVHD development.¹⁶ 

Pre-HSCT conditioning frequently leads to poor oral intake, and nutrition support is routinely required. Increasingly, the literature supports the use of enteral nutrition (EN) over parenteral nutrition (PN), with EN associated with fewer central line complications, reduced need for antifungal therapy,¹⁷ faster immune recovery,^18,19 and reduced risk of GVHD and improved survival.^18,20,21 However, there is no consensus in the literature on the optimal EN formula for patients after HSCT, with provision of elemental,²² semielemental,²³ and polymeric formulas^17,18,24 reported. Few studies have assessed the impact of nutrition support on the microbiome of adults after HSCT. Our work demonstrated higher abundance of taxa associated with SCFA production in patients who received predominantly nonfiber EN compared with PN.²⁵ In addition, patients who maintained a higher oral intake during provision of nutrition support had higher microbial diversity 1 month after HSCT.²⁵ 

Four small pilot studies^26-29 have tested use of an oral prebiotic fiber supplement during HSCT, with some reporting higher butyrate levels²⁹ and improved clinical outcomes.^27,28 However, the amount of fiber consumed was frequently less than prescribed. The side effects of conditioning limit oral intake during HSCT; therefore, provision of a prebiotic fiber–containing EN formula may circumvent some of these issues. In other patient groups, fiber-containing EN had a positive effect on fecal SCFA levels^30-32 and beneficial Bifidobacterium abundance compared with a fiber-free formula^32,33; however, this has not been evaluated in adult HSCT practice.³⁴ This study compared EN tolerance, microbial diversity and composition, SCFA levels, and clinical outcomes in patients receiving standard nonfiber EN vs prebiotic fiber–containing EN after allogeneic HSCT.

This study was approved by the Royal Brisbane and Women's Hospital (RBWH) human research ethics committee (HREC/2021/QRBW/74749) and registered with the Australian New Zealand Clinical Trials Registry (identifier: ACTRN12621000832875). Eligible patients were aged ≥18 years, undergoing myeloablative or reduced intensity myeloablative allogeneic HSCT at RBWH, and provided written informed consent before conditioning. Patients were ineligible if they had received nonmyeloablative conditioning, were enrolled on a GVHD prevention study, had hepatitis B or C or HIV infection, or had an intolerance or allergy to an ingredient in the enteral feed. It was a pilot feasibility study with 10 patients receiving standard care, a nonfiber EN formula; and 20 patients receiving a prebiotic fiber–containing EN formula. The first 10 participants who provided consent were allocated to the standard-care arm and the subsequent 20 patients to the intervention arm. One patient in the control group did not progress to transplant and was replaced after recruitment of the intervention group.

The standard-care arm commenced EN as per the RBWH allogeneic HSCT nutrition support protocol (usual care).³⁵ This included placement of a nasogastric tube and provision of a polymeric nonfiber ready-to-hang formula (1.2 kcal/mL, 63 g protein per liter) on day +1.³⁵ EN was commenced at 30 mL/h of continuous feeding, increasing to 50 mL/h when oral intake declined to under half of meals, and increased to goal rate (providing 100% of estimated requirements) if oral intake was minimal. If EN ceased because of intolerance and oral intake was poor and not expected to improve for a further week, PN was commenced. Both EN and PN ceased once oral intake improved after neutrophil engraftment. During provision of nutrition support, patients were provided with high protein high energy meals, drinks, and snacks, and encouraged to continue oral intake if tolerated. Participants who were allocated to the prebiotic intervention arm increased the feeding rate as per usual care described earlier. In place of the nonfiber EN formula, participants received a polymeric ready-to-hang formula containing inulin, oligofructose, arabic gum, soy polysaccharides, cellulose, and resistant starch (1.2 kcal/mL, 63 g protein, 15 g fiber per liter). If tolerance of EN was poor, participants changed to PN as per standard care described earlier.

The primary end point was tolerance of prebiotic EN, defined as the proportion of patients who did not require cessation of EN and commencement of PN as per the RBWH unit nutrition support protocol. Secondary end points evaluated included amount of prescribed EN received, duration of nutrition support, nutritional status on admission, incidence of infections during hospital admission, length of hospital stay from transplant and incidence of GVHD, disease relapse, and mortality at 100 days after HSCT. Additional end points assessed at 2 time points were stool and blood SCFA levels (after conditioning/before feeding and after feeding), stool microbiome shotgun metagenomic sequencing (baseline/before conditioning and after feeding), and serum cytokines (baseline/before conditioning and after feeding).

Data on oral energy, protein, and fiber consumption were collected for 3 days before conditioning (before first stool sample collection) and daily while inpatients between day −2 and the time of final stool sample collection. Food intake was collected either in the Easy Diet Diary mobile app (version 6.0.28; Xyris Pty Ltd, Brisbane, Australia, 2020) or on a paper diary. Oral intake was assessed for energy, protein, and fiber content through FoodWorks version 10 Professional (Xyris Pty Ltd, 2019) and Delegate software (version 15.10; Delegate Technology, Vienna, Austria, 2021). Nutritional requirements were estimated at 100 to 125 kJ/kg and 1 to 1.5 g/kg protein per day.³⁶ The amount of EN and PN provided daily was collected from fluid balance charts.

The incidence of infections (grade ≥2) and diarrhea was recorded using the Common Terminology Criteria for Adverse Events version 5. GVHD was graded as per standard care³⁷ and mucositis grade using the World Health Organization criteria.³⁸ Nutritional status was assessed using the subjective global assessment. Demographics and clinical outcomes including the incidence of GVHD, antibiotic use, disease relapse, and mortality were collected from medical charts. Standard GVHD prophylaxis was given to all patients, which included cyclosporin or tacrolimus in addition to methotrexate (days +1, +3, +6 and +11 after transplant) or mycophenolate, or posttransplant cyclophosphamide. Antibiotic provision above routine prophylaxis of sulfamethoxazole and trimethoprim 800/160 mg twice a day during conditioning plus ciprofloxacin 500 mg twice a day from day −1 until discharge, or commencement of intravenous β-lactam antibiotics, was recorded.

For both groups, stool samples were collected before conditioning, after conditioning/before feeding and at day +12 to +18 after transplant (after feeding/periengraftment). Stool samples for SCFA analysis (after conditioning and after feeding) were frozen at −20°C within 10 minutes of collection. Serum was separated by centrifugation within 4 to 6 hours of collection and frozen at −80°C in 500 μL aliquots in cryovials. Details of the microbiome, SCFA, and cytokine analysis are included in supplemental methods 1.

Standard descriptive statistics were used to describe the patient cohort and characteristics. To evaluate for differences in baseline characteristics between the prebiotic and standard EN groups, categorical variables were assessed using the χ² or Fisher exact test, and continuous variables using the independent samples t test, or Mann-Whitney U test when data were not normally distributed. Intention-to-treat analysis was completed with statistical significance set at P <.05. Microsoft Excel, Prism version 9.4.0 (GraphPad Software, Boston, MA) and SPSS (released 2022, IBM SPSS Statistics for Windows, version 29; IBM Corp, Armonk, NY) were used.

Biostatistical analysis was completed by Microba. Comparative analyses were performed using R software (version 3.4; R core team, R Foundation for Statistical Computing, Vienna, Austria). An intention-to-treat approach was used. Data transformation with centered log-ratio (clr) transformations was completed before ordination and univariate statistical analysis.

To compare the transformed relative abundances, a linear mixed-effect regression (LMER) analysis was used to compare the differences in microbiome changes between the standard-care and intervention arms over time. Changes within each arm from preconditioning to after feeding were assessed with Welch paired 2-sample t test analysis without adjustment, and an LMER with covariate adjustment. Comparisons of the standard and intervention arms after feeding were analyzed with the Welch 2-sample t test without adjustment, and analysis of variance (ANOVA) with covariate adjustment. Taxon prevalence was compared after feeding with the Fisher exact test or Cochran-Mantel-Haenszel test when adjusting for a covariate. α-Diversity was reported using Shannon diversity and compared between groups using the aforementioned methods. P values were corrected for multiple testing using the Benjamini-Hochberg procedure. Antibiotic use preconditioning was included as a covariate in all analysis. Differentially abundant microbial functions were identified using the univariate methods ANOVA or LMER on clr-transformed relative abundance data, the Fisher exact test, and Aldex2 (ANOVA-like differential expression). Aldex2 was run on read count data. The Fisher exact test was used to test for differences in the presence and absence of microbial functions across study groups.

Unsupervised ordination including nonmetric multidimensional scaling, principal component analysis, principal coordinate analysis, and uniform manifold approximation and projection were used to visualize relationships between samples in 2 dimensions. Nonmetric multidimensional scaling analysis (between groups after feeding) and principal coordinate analysis were performed using Bray-Curtis dissimilarity matrices. Supervised ordination was used to test for relationships between groups after feeding. This included redundancy analysis performed on square root relative abundance and permutational multivariate ANOVA using distance matrices performed using Bray-Curtis dissimilarity matrices. Sparse partial least squares discriminant analysis was performed using clr-transformed relative abundances.

Thirty participants were recruited between October 2021 and May 2022, with 10 patients allocated to the standard-care arm and 20 to the prebiotic EN intervention arm. No significant differences in baseline characteristics were found between groups (Table 1). The nutrition outcomes of both groups are outlined in Table 2. Because of each patient requiring a different feeding rate to meet their estimated requirements in addition to food intake, the dose of fiber received varied widely between 11 and 29 g/d. After transplant, total fiber intake (oral plus enteral) in the prebiotic group was significantly higher than the standard-care group (4 g vs 22 g; P ≤ 0.001; Figure 1). There was no difference in EN duration and the amount of goal EN received between groups.

One patient from each group did not receive EN because of failed nasogastric tube placement and did not require PN because of maintenance of oral intake. These patients are included in all analyses. There was no difference in EN tolerance between groups, with 20% (n = 2) of the standard EN group and 15% of the prebiotic group (n = 3) requiring PN. Reasons for PN commencement included development of enterocolitis (n = 1), loss of the nasogastric tube and tube replacement being contraindicated (n = 3) or declined (n = 1). There were also no differences found in any clinical outcomes between feeding groups, including infection rates, need for antibiotics, incidence of GVHD, disease relapse, and survival (Table 3). When clinical outcomes were examined based on antibiotic provision, the patients who received piperacillin/tazobactam (piptaz) or meropenem after HSCT had significantly more GVHD at day 100 (67% vs 22%; P = .046), compared with those who did not.

An overall reduction in Shannon microbial diversity was observed in both standard and prebiotic EN groups from before to after transplantation. There was no significant difference between groups in Shannon diversity at either time point, and no significant difference in the change between groups (P = .93; Figure 2A). However, some changes were observed at a genus- and species-specific level. There was a significant difference between groups with respect to the change over time in the abundance of the Lactobacillus_C genus, with abundance increasing in the prebiotic group but remaining unchanged in the standard group (P = .013; Figure 2B). In addition, there was a significant change over time in the relative abundance of Lactobacillus_C rhamnosus, with an increase in the prebiotic group over time but not in the standard-care group (P = .022; Figure 2C). The same trend was also observed for Lactobacillus_C paracasei (P = .026; Figure 2D).

The relative abundance of Faecalicatena gnavus increased in the standard group and declined in the prebiotic group over time (P = .0027; Figure 2E). For the relative abundance of Bifidobacterium there was no significant difference between the groups over time (P = .74). The relative abundance of Bifidobacterium in both groups at the postfeeding time point is outlined in Figure 2F (P = .65). At the postfeeding time point, the principal component analysis plot demonstrated minimal divergence between the prebiotic and standard groups (Figure 3).

Functional analysis of microbial sequences using MetaCyc pathways and groups as well as enzyme commission numbers revealed significant differences in the group receiving prebiotic EN before and after feeding, that were not seen in the standard-care group (supplemental table 2). Notably expression of MetaCyc antibiotic resistance gene pathways, which includes vancomycin resistance, β-lactam resistance, and linezolid resistance, were significantly reduced after prebiotic EN feeding. These changes remained significant when corrected for baseline antibiotic use (Figure 4A). Interestingly, alcohol degradation pathways and fatty acid and lipid biosynthesis pathways were also significantly increased after administration of prebiotic EN (Figure 4B-C).

Patient stool and serum samples were analyzed for SCFA and related metabolite levels. In stool, few changes were seen when comparing SCFA content before and after feeding or between standard and prebiotic enteral feeding groups at the same time point (Figure 5A-E,G-H). However, there were significant changes in the ethanol content of stool. Both groups were similar before feeding; however, there was an increase in ethanol content in the control group when prefeeding and postfeeding time points were compared (mean prefeeding stool ethanol content 1.325 μmol/L vs 4.5 μmol/L after feeding; P = .034) that was not seen in the prebiotic group (mean prefeeding stool ethanol content 3.284 μmol/L vs 1.097 μmol/L after feeding; P = nonsignificant (ns); Figure 5F). There was also a significant increase in the ethanol content of the control group when compared with the prebiotic group at the postfeeding time point (postfeeding control group mean ethanol content 4.5 μmol/L vs postfeeding prebiotic group mean ethanol 1.097 μmol/L; P = .0021; Figure 5F).

In serum, only low levels of SCFAs including acetate, lactate, and isobutyrate were detected (Figure 5I-K), and no recordable levels of other related metabolites were seen. No significant differences between before and after feeding and between standard and prebiotic enteral feeding groups were present.

There were no significant differences seen between standard and prebiotic feed groups, or between pretransplantation and posttransplantation time points for levels of interleukin-2 (IL-2), IL-4, tumor necrosis factor, granulocyte-macrophage colony-stimulating factor, or interferon gamma (Figure 6A-B,E-H). There was an increase seen in the levels of IL-6 in the prebiotic feed group after transplantation (before feeding mean IL-6, 139.1 pg/mL vs after feeding, 559.3 pg/mL; P = .0484; Figure 6C). A reduction in postfeeding IL-10 was seen in the prebiotic group when compared with the control group (control group postfeeding IL-10 mean, 51.56 pg/mL vs prebiotic postfeeding IL-10 mean, 17.04 pg/mL; P = .0023; Figure 6D).

This pilot study demonstrated equivalent tolerance of a prebiotic fiber EN formula compared with a fiber-free formula after HSCT. Use of the prebiotic fiber EN did not result in any differences in clinical outcomes including rates of gastrointestinal side effects, PN requirement, or GVHD. Although overall Shannon diversity of the microbiome was not significantly altered by the provision of prebiotic fiber EN, some microbial changes were observed with differences in the relative abundance of Lactobacillus_C rhamnosus, Lactobacillus_C species, Lactobacillus_C paracasei, and F gnavus between groups. In addition, metabolomic profiles and microbial functional pathways were altered, with recipients of prebiotic EN avoiding accumulation of ethanol in the stool observed in the standard-care arm, and recipients of prebiotic EN expressing lower levels of antibiotic resistance genes after feeding. Alterations in systemic serum cytokines were seen, confirming the impact of a brief, fiber-containing dietary intervention on the gut microbiome–metabolome-immunity axis, and conferring functional microbiome changes.

Several studies have evaluated tolerance of EN during HSCT, with levels varying and reported up to 76%.^17-19,24 The variation in EN tolerance is likely due to many factors including timing of commencement, feeding rate, management of HSCT side effects such as nausea and vomiting, adequate staff training and support for EN, and early patient education.³⁹ Although recent research has focused on outcomes of provision of EN vs PN, research investigating the role of nutrition to support the gastrointestinal microbiome, and its impact on the metabolome and immune function during HSCT has been limited. To our knowledge, no studies have evaluated fiber-containing EN for adults undergoing hematological cancer treatments.³⁴ Previous studies have investigated provision of an oral prebiotic fiber supplement during HSCT.^26-29 Two reported improved clinical outcomes, including reduced duration of severe mucositis and incidence of acute GVHD,²⁸ and lower mortality (with prebiotic plus Lactobacillus probiotic).²⁷ In addition, higher abundance of butyrate producing bacteria²⁸ and higher fecal butyrate levels with prebiotic supplementation have been reported.²⁹ 

The prebiotic type and dose provided in these studies varied widely; 1 to 3 fibers were tested with doses between 5 and 40 g/d.^26-29 The EN formula in this study contained 6 fiber types. Provision of fiber-containing EN ensures consistent fiber provision during the neutropenic period when oral intake is low. In this study, participants consumed a median of 4 g fiber per day from food intake in the standard-care arm after transplant. Although the intervention group’s overall fiber intake improved to an average 22 g/d, this remains below recommendations for fiber intake,⁴⁰ providing rationale for testing of higher fiber doses and combinations for tolerance and benefit. Furthermore, the average oral fiber intake was higher in the prebiotic group after HSCT but not significantly different between groups. Although the literature on fiber supplementation during HSCT does not yet indicate an optimal fiber dose or type, in healthy populations a varied high-fiber diet has been associated with higher microbial diversity,¹³ indicating that prebiotic supplementation containing a variety of fiber types requires evaluation.

In this study, the prebiotic group had higher relative abundance of Lactobacillus_C rhamnosus and the Lactobacillus_C genus. In addition, after feeding the relative abundance of Lactobacillus_C paracasei was higher in the prebiotic group than in the standard group. Lactobacillus species including Lactobacillus rhamnosus are known to be associated with protecting the gastrointestinal barrier.^41,Lactobacillus rhamnosus can have an antipathogenic effect and reduce proinflammatory cytokines.⁴² The role of Lactobacillus including Lactobacillus rhamnosus and Lactobacillus paracasei strains in producing SCFAs including butyrate has also been observed.⁴² However, the role of Lactobacillus during HSCT is still unclear. In preclinical studies, Lactobacillus rhamnosus supplementation reduced gut inflammation and the incidence of GVHD and mortality⁴³; however, this was not observed in clinical trials.^44,Lactobacillus paracasei abundance has been associated with acute GVHD in 1 study.⁴⁵ However, supplementation of Lactobacillus acidophilus demonstrated anti-inflammatory and immunomodulatory properties and may reduce GVHD severity.⁴⁶ In pediatric HSCT, high abundance of Lactobacillus species is also positively correlated with T-cell reconstitution.⁴⁷ 

Increased Lactobacillus abundance after prebiotic supplementation has been reported in several studies evaluating provision of inulin or fructo-oligosaccharide in healthy populations and other cancer treatments.^48-51 However, inulin-type fructan supplementation (oral or EN) is also associated with an increase in the relative abundance of beneficial Bifidobacterium SCFA-producing species.^32,33,50 This finding was not observed in this study and no difference in microbial diversity or levels of SCFAs was observed, despite an increase in fatty acid and lipid biosynthesis pathways. This may reflect the multifactorial insults to the microbiome that occur during HSCT.¹⁰ Reduced microbial diversity can occur before HSCT,⁵² before our intervention. Stool SCFA levels are also a poor measure of SCFA production.¹⁴ Our results may also reflect the frequent provision of broad-spectrum antibiotics or indicate a relatively short and potentially inadequate dose of prebiotic fiber. In future studies, a longer intervention should be considered to test protection of the microbiome from conditioning commencement. Early oral fiber supplementation in addition to an enteral fiber supplement may be required to optimize fiber intake across the transplant course.

In this study, relative abundance of F gnavus (also referred to as Mediterraneibacter gnavus and Ruminococcus_B gnavus) increased in the standard group and declined in the prebiotic group after feeding. F gnavus is an ethanol-producing, mucus-using species that has been associated with nonalcoholic steatohepatitis and inflammatory bowel disease.^53-55 It has also been associated with alterations in immune reconstitution after bone marrow transplantation⁵⁶ and may increase in abundance on a fiber-free diet.⁵⁷ This is consistent with our finding of higher fecal ethanol levels in the standard-care group and increased functional microbial alcohol degradation pathways in the prebiotic fiber group. Ethanol can induce gastrointestinal barrier dysfunction,⁵⁸ which exists after conditioning.⁵⁹ Damage to the gastrointestinal barrier and an increase in intestinal permeability has been associated with the development of bloodstream infections⁶⁰ and GVHD.⁵⁹ 

Importantly, despite the modest changes seen in microbial species diversity and abundance, potentially clinically relevant microbial functions were altered after prebiotic feeding. The gut microbiome is recognized to function as a reservoir for the so-called “resistome”^11,61 and colonization of HSCT recipients with multidrug resistant organisms is associated with an increased risk of infection and higher transplant-related mortality.⁶² In cases in which approaches to reduce the pretransplant burden of multidrug resistant colonization such as decontamination using nonabsorbable antibiotics have been ineffective in preventing clinical infection,^63,64 and approaches such as fecal microbial transplantation carry risk in HSCT recipients,⁶⁵ dietary supplementation with prebiotic fiber may offer a broadly applicable and feasible alternative to reduce the burden of antibiotic resistance gene expression and, potentially, clinical infection.

We observed an impact of microbial species and stool metabolites after prebiotic EN on systemic immunity as reflected in serum cytokines. Here, our findings are notable because they differ from findings of other groups. Increased serum IL-6 is associated with adverse clinical outcomes,⁶⁶ however, IL-6 blockade does not profoundly reduce rates of acute GVHD clinically.⁶⁷ In our study, the alterations in serum IL-6 were not associated with any changes in rates of acute GVHD. Similarly, we observed a reduction IL-10 in the prebiotic group after feeding compared with the control group, without an impact on clinical outcomes, noting the low rate of acute gastrointestinal GVHD in the whole cohort. Cytokines were only measured at a single time point after transplant, however, and may not reflect temporal kinetics of microbiome-linked cytokine changes. We did however see a higher rate of GVHD at day 100 in patients who had received piptaz (piperacillin and tazobactam) or meropenem, which has been reported in other trials.^68,69 

This study is limited by sample size, short duration of the fiber intervention, the lack of baseline SCFA levels, and the lack of phenotyping of immune cell reconstitution after transplant, which were design limitations of this pilot. The higher rate of diarrhea and gut GVHD in the prebiotic fiber group, although not statistically significant, reinforces the need for further evaluation of these outcomes in a larger sample size. In this trial there were also some incomplete food diaries, precluding analysis of food intake for every patient. However, the data collected on oral dietary intake are a strength of this work, in that, to our knowledge, no other studies have reported routine fiber intake before or during adult HSCT. This study highlights the extremely low dietary fiber intake consumed by patients undergoing HSCT.

This pilot trial demonstrates that a prebiotic EN formula is equivalent in tolerability to a fiber-free formula for adult patients undergoing HSCT. Provision of prebiotic EN is associated with increased abundance of some SCFA-producing Lactobacillus species and changes in F gnavus abundance, a reduction in expression of antibiotic resistance genes, and with associated changes in stool metabolite content and serum cytokine levels. Our findings add to the literature exploring the dynamics and therapeutic potential of the gut microbiome–metabolome-immunity axis.

The authors acknowledge the contribution of Siok Tey of the QIMR Berghofer Medical Research Institute for storing the study samples. The Centre for Advanced Imaging at The University of Queensland is acknowledged for access to their nuclear magnetic resonance instruments.

This research was supported by a grant from the Nutricia Research Foundation and a Collaborative for Allied Health Research, Learning, and Innovation Early Career Fellowship (S.A.) and a Metro North Clinician Research Fellowship (A.H.).

No intellectual or editorial input was provided from the funding bodies.

Contribution: S.A. carried out the study design, data collection, analysis of baseline and clinical outcomes, and manuscript drafting; A.H. contributed to data collection, study design, data analysis, and manuscript writing; G.K., M.B., and B.F. contributed to study design and manuscript writing; and all authors have read and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Sarah Andersen, Dietetics and Foodservices, Royal Brisbane and Women’s Hospital, Butterfield St, Herston, QLD 4029, Australia; email: sarah.andersen@health.qld.gov.au.