Systematic review and meta-analysis of interventions with dietary supplements, including pre-, pro- and synbiotics, to reduce acute and late gastrointestinal side effects in patients undergoing pelvic radiotherapy

Objective: Pelvic radiotherapy (RT) often results in toxicity to the gastrointestinal tract and clinical trials have demonstrated a potentially beneficial role of dietary supplements in overcoming this problem, mainly acute effects. This systematic review and meta-analysis were undertaken with the aim of evaluating the efficacy of several dietary supplement interventions in preventing or alleviating symptoms of gastrointestinal toxicity in patients undergoing RT for a range of common pelvic malignancies. Design: The search protocol was prospectively submitted to PROSPERO at the University of York. CENTRAL, MEDLINE, EMBASE, and ClinicalTrials.gov were searched up to June 2020 for randomised controlled clinical trials. Interventions included four supplement categories: biotics, amino acids, poly-unsaturated fatty acids and polyphenols. Efficacy was determined with reference to outcomes based on symptoms of acute gastrointestinal toxicity, including diarrhoea, nausea and vomiting. Results: Twenty-three randomised controlled trials (1919 patients) were identified in this review. Compared with placebo, probiotics, synbiotics and polyphenols were significantly associated with a lower risk of diarrhoea. Biotic supplements also reduced the risk of moderate to severe diarrhoea and the need for anti-diarrhoeal medication. In contrast, amino acid supplements had no effect on acute symptoms. There was a non-significant trend for reduction in nausea and mean bowel movements per day using dietary supplements. Conclusions: Biotic supplements, especially probiotics and synbiotics, reduce acute symptoms of gastrointestinal toxicity in patients undergoing pelvic radiotherapy. More randomised placebo-controlled studies with larger numbers of patients are needed to define specific formulations for supplement interventions that will be of greatest benefit to patients.


BACKGROUND
Radiotherapy is a major cancer treatment modality, used to treat approximately 50% of 2 patients 1 . Chemoradiation (concurrent delivery of systemic chemotherapy) is generally 3 preferred over radiotherapy alone for most pelvic malignancies, including tumours of the 4 lower gastrointestinal, gynaecological and urological (with the exception of prostate cancer) 5 tracts 2 . Over 200,000 patients in the US are treated with pelvic or abdominal radiotherapy 6 each year 3 . It is inevitable that normal gastrointestinal tissues are exposed to radiation 7 during pelvic radiotherapy 4 , with approximately 80% of patients developing acute 8 symptoms of radiation-induced gastrointestinal toxicity 5 . However, despite their impact on 9 patients' quality of life, no prophylactic agents for the alleviation of gastrointestinal side-10 effects from pelvic radiation have been approved to date 6 . 11

12
The intestine is particularly radiosensitive due to the presence of rapidly replicating crypt 13 cells. Following irradiation, mucosal biopsies show evidence of extensive surface epithelial 14 atrophy, widespread disruption of absorptive and secretory activity and acute crypt 15 inflammation crypts 7 . Acute symptoms usually develop during or immediately after RT, and 16 typically improve within three months following RT 8 . The most common acute side effect is 17 diarrhoea, affecting up to 80% of all patients 9 . Other symptoms, such as abnormal stool 18 output, vomiting and gastrointestinal discomfort are also very common. Late symptoms 19 include GI bleeding, fistula, stricture and colostomy 10 . 20 21 Use of a dietary supplement is aimed at boosting daily intake of specific nutrients, to much 22 higher levels than obtained from the diet, to alleviate symptoms of gastrointestinal toxicity. 23 Such dietary supplements include biotics, amino acids, poly-unsaturated fatty acids (PUFAs) 24 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint and polyphenols. Probiotics, mainly of the Lactobacillus and Bifidobacteria genera, are live 1 microorganisms thought to produce health benefits following passage to the intestine 11  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint bean are similarly implicated in anti-inflammatory responses, protecting irradiated mouse 1 jejunum from adverse changes and attenuating tumour growth 19 . 2 3 This review tests the hypothesis that administration of oral dietary supplements for cancer 4 patients receiving pelvic radiotherapy may trigger changes in the lower gastrointestinal tract 5 which lead to a reduction in gastrointestinal toxicity. We conducted a systematic review and 6 meta-analysis of randomised controlled trials (RCTs), with the aim of determining whether 7 dietary interventions using supplements can alleviate symptoms of gastrointestinal toxicity 8 in pelvic RT patients, including diarrhoea, nausea, vomiting, flatulence/bloating and bowel 9 movement frequency. 10 11 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. The full set of search strategies is available in Appendix 1 to 3, and inclusion and exclusion 8 criteria are available in the PROSPERO registration 20 . Relevant articles were identified on 9 PubMed. Handsearching of meta-analyses, systematic reviews and papers identified studies 10 not indexed in the electronic databases used for this review. All titles and abstracts 11 retrieved by electronic searches were downloaded and duplications removed using EndNote 12 reference management software. Full-text copies of short-listed titles were obtained to 13 determine if they met the eligibility criteria. Multiple reports of the same study data were 14 eliminated, thus any original data were only included once. 15

Data extraction 16
Systematic data collection from included studies was conducted using a data collection form 17 designed specifically for this review. It included the following information (where available) 18 for each dataset: publication year, study design, participants (number, age distribution, 19 gender distribution, details of malignancy, details relevant to inclusion and exclusion 20 criteria), current cancer treatment (duration, fractionation and dose of RT and details of any 21 chemotherapy), intervention (type of supplement, route of intervention, formulation, dose, 22 timing with respect to RT, length of treatment and follow-up) and measured outcomes 23 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint (diarrhoea incidence and severity and details of other GI symptoms recorded as toxicity 1 scores, patient-reported outcome measure scores, questionnaires or interview scores). 2 Details of included studies were summarised according to the PICO model 21 . Where raw 3 data were unavailable, EndNote was used to facilitate extraction of data from charts and 4 graphs of the included studies 22 . Details of outcomes included time points at which they 5 were reported by study authors. 6

Outcome assessment 7
Different measures of treatment effects were used for dichotomous and continuous 8 outcomes, namely, risk ratio (RR) for dichotomous outcomes and the mean difference (MD) 9 between the intervention and control arms for continuous outcomes. 10 Standardised mean difference (SMD) was used to compare to compare results from studies 11 that reported the same outcomes measured on different scales. 12

Study quality 13
Risk of bias assessment was carried out for all studies that met the inclusion criteria, using 14 the Cochrane Risk of Bias 2 (ROB2) tool 23 . The ROB2 framework investigates five domains in 15 a study, namely, the biases arising from the randomisation process, due to deviations from 16 intended interventions, due to missing outcome data, in measurement of the outcome and 17 in the selection of the reported result, with three levels of risk of bias, namely, low risk, 18 some concerns and high risk. The risk of bias across the five domains was then used to 19 derive an overall risk score. Judgements were based on assessment of statements in the 20 included studies relevant to study design. In some instances, key information was 21 incomplete, unclear or omitted from published protocols. 22

Assessment of heterogeneity and publication bias 23
. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint To assess the heterogeneity, we used a chi-squared test and I 2 24 25 . P values less than 0.1 1 were considered as evidence of heterogeneity. I 2 values between 0% to 40% indicated 2 'might not be important', 30% to 60% indicated 'may represent moderate heterogeneity', 3 50% to 90% indicated 'may represent substantial heterogeneity' and 75% to 100% indicated 4 'considerable heterogeneity' based on the Cochrane handbook 26 . Tau-squared is the 5 estimated standard deviation of underlying effects across studies. Begg's funnel plots were 6 used to visually assess asymmetry potentially due to publication bias. 7

Data synthesis and statistical analysis 8
Meta-analyses were performed to measure the effect of dietary supplements on an 9 outcome, in instances where there were three or more studies that reported the same 10 outcome. All analyses were conducted using RevMan 5.4 and R version 4.0.2 with package 11 'meta' 27 28 . For dichotomous outcomes, risk ratios (RR) were estimated and were meta-12 analysed using a random effects model using the Mantel-Haenszel (M-H) method 29 . For 13 continuous outcomes, mean differences (MD) were estimated and were pooled using a 14 random effects model with the inverse variance (IV) method. 95% confidence intervals (CI) 15 for all estimates were calculated. For each outcome, the studies selected for quantitative 16 analysis were divided into subgroups, based on the class of intervention. Studies that met 17 inclusion criteria investigated the effect of one of the following classes: biotics, amino acids, 18 PUFAs and polyphenols. Subgroup analysis by subclass of biotic intervention (prebiotic, 19 probiotic and synbiotic) was also performed. Meta-regression by mean age, proportion of 20 male participants and sample size was used to assess whether the effects of interventions 21 on incidence of diarrhoea varied by these study characteristics. 22

Patient and public involvement 23
There was no patient or public involvement in the design or conduct of this study. 24 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. .

1
The search of the four primary databases identified 23,542 titles published between 1946 2 and June 2020 (search process summarised in Figure 1). After 5,825 duplications were 3 removed, a total of 17,717 entries remained. These studies were manually reviewed by title 4 and abstract and 17 met the inclusion criteria. Six further studies were identified from 5 manual searches of the reference sections of research articles. Finally, 23 studies met the 6 inclusion criteria and could be used for quantitative analysis. The effects of interventions on 7 incidence of diarrhoea did not vary by mean age, proportion of male participants or sample 8 size ( Figure S1). Results of the overall and individual risk of bias assessments for each of the 9 five domains are presented in Figure S2. Fourteen studies had low risk, three studies had 10 moderate risk, and six studies had high overall risk of bias. 11 12

Included studies and characteristics of included studies 13
In total 23 studies involving a total of 1,919 patients met the inclusion criteria and for each 14 outcome, they were grouped by intervention category. These studies were all randomised 15 controlled trials and their characteristics are shown in the Table 1. All included studies 16 investigated the effect of dietary supplements on acute symptoms of gastrointestinal 17 toxicity. However, the individual trials differed widely in the symptoms and outcomes they 18 reported. In total, the trials included in the review reported ten different relevant 19 symptoms. For each symptom, data could be expressed according to the following three 20 outcomes: number of events, severity and time, as shown in Table S1. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint

Efficacy of dietary supplements in preventing diarrhoea 1
The meta-analysis comprising 1,625 patients showed that dietary supplements reduced the 2 risk of diarrhoea ( Figure 2). The pooled risk ratio (RR) was 0.79 (95% CI: 0.66 to 0.94; 3 P=0.007) and there was significant heterogeneity amongst the studies (I 2 =73%; P<0.001). 4 The funnel plot for this meta-analysis ( Figure 3) was largely symmetrical, implying only mild 5 publication bias in the references included. There was no evidence that heterogeneity was 6 due to mean age or sex of participants or sample size of the studies. Meta-analyses were 7 then carried out for biotic, amino acid, poly-unsaturated fatty acid and polyphenol 8 interventions. Amino acid interventions were not associated with risk of diarrhoea with a RR of 1.05 (95% 21 CI=0.86 to 1.29; P=0.65, Figure 2). Five studies which included 348 patients were used to 22 compare the incidence of diarrhoea between intervention and control groups. We found 23 that four studies had consistent results of RR which were close to 1, but only Vidal-24 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. There was only one study in the category of poly-unsaturated fatty acids reporting the 5 outcome incidence of diarrhoea ( Figure 2). This was a relatively small randomised controlled 6 trial with only 40 patients. This study had a RR of 0.63 (95% CI=0.25 to 1.58; P=0.32). 7 8 Efficacy of polyphenol in preventing diarrhoea 9 Two studies compared polyphenols and placebo among 64 patients ( Figure 2). Both showed 10 that the intervention was associated with lower incidence of diarrhoea. The overall RR was 11 0.30 (95% CI=0.13 to 0.70, P=0.005). There was no evidence of heterogeneity between 12 these two studies (I 2 = 0%, P=0.86). 13 14 .
CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint Efficacy of dietary supplements in preventing moderate to severe diarrhoea 1 Efficacy of dietary supplements was assessed against moderate to severe diarrhoea, with 2 the incidence of moderate to severe diarrhoea defined as the incidence of grade 2 or higher 3 diarrhoea, based on Common Technology Criteria for Adverse Events (CTCAE) 31 ( Figure 5). 4 Dietary supplements were associated with a lower risk of moderate to severe diarrhoea 5 (RR=0.68; 95% CI: 0.48 to 0.98; P=0.04). There was high heterogeneity among studies 6 (I 2 =65%, P=0.009). The subgroup analysis suggested that the association was mainly driven 7 by biotic interventions for which the RR was 0.49 (95% CI: 0.36 to 0.67; P<0.001), but not 8 amino acids (RR=1.05; 95% CI: 0.82 to 1.34; P=0.70). 9 10

Efficacy of dietary supplements in preventing the use of anti-diarrhoeal medication 11
Anti-diarrhoeal medication, such as loperamide, is often employed for patients who 12 experience diarrhoea during or after radiotherapy. Therefore, we measured the effect of 13 dietary supplements against the incidence of anti-diarrhoeal medication use ( is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. results also showed that the interventions had no effect on vomiting and flatulence/bloating 7 with relative risks of 0.99 (95% CI: 0.79 to 1.25, P=0.95) and 1.12 (95% CI: 0.59 to 2.12; 8 P=0.72) respectively. 9 10 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint DISCUSSION 1 In this review, 23 randomised controlled trials met inclusion criteria for quantitative 2 analysis. Risk of bias assessment was conducted for each of these studies. Meta-analyses 3 were carried out for seven of these outcomes. These showed that dietary supplements are 4 effective in reducing the risk of diarrhoea, experiencing moderate to severe diarrhoea and 5 anti-diarrhoeal medication use. Subgroup analysis showed that biotic supplements and 6 polyphenols were effective in reducing the risk of these outcomes, but amino acids were 7 ineffective. Among the subclasses of biotic interventions, both probiotic and synbiotic 8 supplements were shown to be effective in reducing the risk of diarrhoea. Taken together, 9 these results indicate that biotic supplements can reduce the risk of acute diarrhoea and the 10 severity of diarrhoea in patients undergoing pelvic radiotherapy. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. analysed. Therefore, the inclusion of prebiotics in this review is another key distinguishing 20 feature. The previous meta-analyses investigating the effects of biotic supplements on acute 21 symptoms of gastrointestinal toxicity are listed in Table S2. Overall, the trials included in each 22 meta-analysis are largely identical. Where different selections were made, this is mainly due 23 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint to variations in search strategy and the statistical methods used. A Cochrane systematic 1 review has also investigated the efficacy of interventions, including radiotherapy techniques 2 and pharmacological and non-pharmacological interventions, on acute and late adverse 3 gastrointestinal effects of pelvic radiotherapy for primary pelvic cancers 4 . Compared to their 4 study of non-pharmacological interventions, including dietary interventions, probiotics, 5 glutamine, counselling, and protein supplements, our focused approach showed that the 6 beneficial effects mainly came from the probiotics and synbiotics. Also, the focus of the 7 Cochrane review was on prevention, rather than treatment, of acute symptoms of 8 gastrointestinal toxicity, an important difference to our review. Additionally, their search was 9 only updated to November 2017; our search to June 2020 included three more recent studies is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint toxicity, rather than less direct measures of effect, such as immunological, biochemical or 1 histological markers. Therefore, the findings of this review can be directly applied to clinical 2 practice. However, patient-related factors of low numbers (<40) of participants in 26% of 3 included studies represent a weakness in the statistical treatment of data sets, resulting in 4 wider confidence intervals. Most data were derived from single-centre trials, thus limiting the 5 ability to recruit larger numbers of patients. Furthermore, the baseline heterogeneity of the 6 patient cohorts is high, as it is not possible to subdivide patients into meaningfully 7 representative sub-groups in terms of age, sex, lifestyle-related factors, co-morbidities and 8 type of cancer. 9 The underlying protective effects of dietary supplements against GI toxicities may be 10 mediated as shown in Figure 7. A direct effect on the intestinal immune environment 11 following intake of specific dietary agents may lead to anti-inflammatory changes that 12 alleviate gastrointestinal toxicity. There may also be an indirect effect, whereby the above 13 immunomodulatory actions are developed in response to changes in the gut microbiota and 14 their metabolites, particularly SCFAs. 15 In terms of probiotics, these can act in three beneficial ways. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . pathophysiology of chronic effects is more complicated than acute effects 48 . Evidence from 1 clinical studies suggests that acute and chronic effects are linked, with the risk of developing 2 late effects greater in patients that have developed acute effects (consequential late 3 effects) 49-51 . Consistent with this finding, a systematic review that investigated the 4 relationship between acute and late gastrointestinal toxicity after RT for prostate cancer 5 concluded that acute GI toxicity may be predictive of high risk of developing late effects 51 . 6 Taken together, the evidence suggests that dietary supplements could ameliorate chronic 7 symptoms of gastrointestinal toxicity in two ways: directly, by minimising the 8 pathophysiology underlying chronic effects, including differentiation of mitotic fibroblasts 9 into post-mitotic fibroblasts, and indirectly, by reducing the magnitude of acute symptoms 10 and the resulting consequential late effects. 11 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint CONCLUSION 1 Findings from our systematic review and meta-analysis suggest that biotic supplements, 2 specifically probiotics and synbiotics, are effective in reducing the risk and severity of acute 3 symptoms of gastrointestinal toxicity caused by pelvic radiotherapy. The data also broadly 4 confirm that they are safe to use. 5 6 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020.

Competing interests 16
The authors declare that they have no competing interests. 17 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Ethics Approval and Consent to Participate
The copyright holder for this this version posted August 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted August 23, 2020. . https://doi.org/10.1101/2020.08.21.20178814 doi: medRxiv preprint