Introduction
In the United Kingdom, almost four in every ten patients with cancer receive radiotherapy as part of their treatment.Reference Williams, Summers, Drinkwater and Barrett1 When the patient's body volume treated includes part or all of the gastrointestinal tract, a common side effect is diarrhoea.Reference Blanarova, Galovicova and Petrasova2
Radiotherapy-induced diarrhoea (RID) is defined as an inflammatory and degenerative process caused by radiation and affecting the gastrointestinal tract.Reference Blanarova, Galovicova and Petrasova2 Usually, diarrhoea starts within the first 2 weeks of treatment; it can be mild and eventually resolve, or can become more serious and chronic.Reference Bismar and Sinicrope3 In the United Kingdom, ∼80% of patients with cancer undergoing abdominal or pelvic radiotherapy will develop some degree of RID, resulting in patients’ distress and potential withdrawal of treatment.Reference Andreyev4
Radiation is thought to cause diarrhoea by creating changes in intestinal bacterial flora, intestinal motility, vascular permeability of the mucosal cells and intestinal mucosa malabsorption of lactose and bile acids.Reference Blanarova, Galovicova and Petrasova2, Reference Salminen, Elomaa, Minkkinen, Vapaatalo and Salminen5 Treatment often includes medications such as antibiotics, sucralfate, mesalazine, octreotide, loperamide and diphenoxylate.Reference Visich and Yeo6, Reference Timko7 While exact statistics are lacking, treatment failure is thought to occur in a large proportion of patients,Reference Muehlbauer, Thorpe, Davis, Drabot, Rawlings and Kiker8 highlighting the need for novel approaches in the prevention of RID.
In the last few years, probiotics have acquired importance in the prevention and treatment of acute gastrointestinal symptoms.Reference Maria-Aggeliki, Nikolaos, Kyrias and Vassilis9–Reference Urbancsek, Kazar, Mezes and Neumann11 A joint report by the Food and Agriculture Organisation of the United Nations and the World Health Organisation defines probiotics as ‘live microorganisms which when administered in adequate amounts confer a health benefit on the host’.12 Probiotics provide their benefit through modulation of intestinal inflammation by altering the composition and the metabolic properties of the indigenous intestinal flora.Reference Timko7 In experimental and clinical studies, probiotics have been shown to be beneficial in the prevention and treatment of a variety of gastrointestinal disorders, including infectious diarrhoea, clostridium difficile-induced diarrhoea, inflammatory bowel disease, irritable bowel syndrome and ulcerative colitis.Reference Famularo, De Simone, Matteuzzi and Pirovano13 This systematic review aims to examine the evidence for a possible efficacy of probiotics in the prevention of RID.
Method
The review follows the National Institute for Health and Clinical Excellence (NICE) guidelines manual methodology,14 and it is designed as a ‘mini-review’, along the lines suggested by Griffiths.Reference Griffiths15 To search for relevant studies, we asked the question: are probiotics more efficacious than placebo at preventing RID in adults with cancer? We searched Medline and Embase databases. Medline has a broad coverage of biomedical literature, whereas Embase has a stronger focus on pharmacology, drug research and European literature.14, Reference Wong, Wilczynski and Haynes16 Only randomised controlled trials (RCTs) were included, as they are the studies of choice when answering questions of therapeutic efficacy.Reference Kaptchuk17, 18Table 1 reports the inclusion/exclusion criteria used.
Table 1 Inclusion and exclusion criteria
Results
The database search provided 63 original citations. We excluded irrelevant papers by scrutinising the title or abstract, or full copy where there was uncertainty as to whether they met the inclusion criteria (Figure 1, Table 2). Finally, four papers answering the review question and meeting the inclusion criteria were selected.Reference Delia, Sansotta and Donato19–Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22
Figure 1 Exclusion of papers.
Table 2 Excluded papers
Abbreviations: RCT, randomised controlled trial; RID, radiotherapy-induced diarrhoea.
Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 conducted a randomised, double-blind, placebo-controlled trial to determine whether a probiotic preparation (Infloran) reduced the risk of RID. They recruited 63 adult female patients (two groups) undergoing radiotherapy and concurrent cisplatin chemotherapy for cervical cancer (they did not specify the health-care settings).
Delia et al.Reference Delia, Sansotta and Donato19 investigated the ability of a high-potency probiotic preparation (VSL#3) to prevent RID by conducting a randomised, double-blind, placebo-controlled trial. They recruited 490 adult patients (two groups) attending an Italian radiotherapy outpatient unit, who were receiving radiotherapy for sigmoid, rectal or cervical cancers.
Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 studied the impact of a standard dose and a high dose of a probiotic preparation (Bifilact) on the onset of RID by conducting a randomised, placebo-controlled trial. They recruited 246 adult patients (three groups) of a hospital in Quebec with rectal, cervical, endometrial or prostatic cancer, with some patients having received surgery or chemotherapy.
Giralt et al.Reference Giralt, Regadera and Verges21 ran a randomised, double-blind, placebo-controlled trial to study the efficacy of a fermented liquid yogurt (containing Lactobacillus casei DN-114001) for the prevention of RID. They recruited 118 adult female patients (two groups) receiving radiotherapy and concurrent cisplatin chemotherapy for cervical or endometrial cancer at different Spanish hospitals.
Quality assessment
As suggested by Huwiler-Muntener et al.,Reference Huwiler-Muntener, Juni, Junker and Egger23 we separately assessed the reporting quality and methodological quality of the four papers. Reporting quality was assessed with the CONSORT RCT checklist.Reference Schulz, Altman and Moher24 The papers did not describe the randomisation method in detail, nor calculated effect sizes and confidence intervals (CIs). The paper by Delia et al.Reference Delia, Sansotta and Donato19 and Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 published, respectively, as ‘rapid communication’ and, as a conference proceeding abstract, had the lowest reporting quality, lacking discussion of blinding, statistical methods and study limitations.
Methodological quality was assessed with the NICE RCT methodology checklist25 combined with the GRADE risk of bias criteria for individual studies.Reference Guyatt, Oxman and Vist26 The assessment aimed at checking for sources of systematic biasReference Greenhalgh27 for the presence of diarrhoea (Table 3). Owing to lack of clarity regarding their selection methods, all four studies showed risk of selection bias. The study by Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 was the only study with no risk of attrition bias, as there were no withdrawals among participants and intention to treat analysis was used.Reference Hollis and Campbell28 The study by Giralt et al.Reference Giralt, Regadera and Verges21 had the highest risk of attrition bias, with 27% attrition ratio and per protocol analysis. Moreover, attrition ratio was imbalanced, with 34% and 21% withdrawals for placebo and treatment group, respectively, increasing risk for bias.Reference Guyatt, Oxman and Vist26 The study by Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 had the least risk of detection bias, as stool consistency was assessed objectively. The study by Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 may be prone to systematic bias, as it lacked clarity in the randomisation process and blinding in the study design, as well as in how diarrhoea was defined (Table 4).
Table 3 Risk of systematic bias for the presence of diarrhoeaFootnote a
Note: a Adapted from Guyatt et al.Reference Guyatt, Oxman and Vist26
Of the four studies, Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 and Delia et al.Reference Delia, Sansotta and Donato19 were the most robust; the study by Giralt et al.Reference Giralt, Regadera and Verges21 was the least robust, and the robustness of the study by Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 could not be estimated.
Findings and discussion
Tables 4 and 5 report features and findings of the studies. Effect sizes and their CIs were calculated as number needed to treat (NNT).Reference Altman29, Reference Newcombe and Altman30 NNT, endorsed in the GRADE system,Reference Guyatt, Oxman and Kunz31 is an absolute measure of effects conveying both statistical and clinical significance.Reference Cook and Sackett32 The NNT was calculated from the absolute risk reduction (ARR), which is also reported (Table 5). When the ARR is negative, which occurs when the treatment has a harmful effect, the NNT is also negative. A positive (beneficial) NNT is indicated as NNTB and a negative (harmful) NNT as NNTH, and both have a positive sign.Reference Altman33
Table 4 Features of included studiesFootnote a
Note: a Adapted from the evidence table for interventions studies in NICE.25
Table 5 Findings of included studies for the presence of diarrhoea and calculated effect sizes
Notes: aStudy findings expressed as risk of diarrhoea, shown as ratio between number of patients with diarrhoea in a group and total number of patients in that group, for both intervention group (I) and control group (C).
bThe number needed to treat (NNT) is calculated as the inverse of the absolute risk reduction (ARR). For each study, ARR was calculated as the risk of diarrhoea in the control group minus the risk of diarrhoea in the treatment group. When the ARR is negative (i.e., treatment is worse than placebo), the NNT is also negative. A positive NNT is indicated as NNTB (benefit) and a negative NNT is indicated as NNTH (harm), both with a positive sign.
cGermain et al.Reference Germain, Desjardins, Demers and Dagnault20 reported only the percentage of participants with diarrhoea in each group; therefore, to estimate the boundaries of the confidence intervals, the groups were assumed to be of equal number of participants, that is, 246/3 = 82.
A beneficial effect of their probiotic preparation was found by Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 (NNTB 2·2, 95% CI 1·5–4·1) and by Delia et al.Reference Delia, Sansotta and Donato19 (NNTB 5, 95% CI 3·5–8·6) and both results were statistically significant. Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 found a beneficial effect for both standard dose group (NNTB 5·5, 95% CI 3·2–19·5), which was statistically significant, and high dose group (NNTB 10·3, 95% CI NNTH 35 to ∞ to NNTB 4·5), which did not reach statistical significance. These results have a high clinical significance: very few patients (the NNTB value) needed to be treated to prevent one additional case of diarrhoea. Probiotics are considered safe, and although adverse events such as probiotic bacteraemias exist they are very rare.Reference Heselmans, Reid, Akkermans, Savelkoul, Timmerman and Rombouts34 Only the study by Giralt et al.Reference Giralt, Regadera and Verges21 found an unfavourable effect of the treatment, which was not statistically significant (NNTH 10·4, 95% CI NNTH 3·3 to ∞ to NNTB 9·3). None of the four studies reported adverse events.
Baseline characteristics of intervention and control groups had no significant differences in each of the three studies that reported relevant details,Reference Delia, Sansotta and Donato19, Reference Giralt, Regadera and Verges21, Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 providing good internal validity. Participants’ total radiation dose was similar across the three studies that reported it.Reference Delia, Sansotta and Donato19, Reference Giralt, Regadera and Verges21, Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 Some of the participants in all studies but those in Delia et al.Reference Delia, Sansotta and Donato19 also received chemotherapy. However, the study samples were considerably different. Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 studied 63 women treated for cervical cancer, and Giralt et al.Reference Giralt, Regadera and Verges21 recruited 118 women with cervical and endometrial cancer. Delia et al.Reference Delia, Sansotta and Donato19 studied 482 adults of both genders treated for various cancers, thereby being more representative of the larger population of patients undergoing abdominal and pelvic radiotherapy, and therefore having higher external validity. Similarly, Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 investigated 246 adults of both genders treated for various cancers.
Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 showed low detection bias. An independent laboratory technician established stool consistency as loose (diarrhoea) and soft or formed (normal stool). However, the much larger studyReference Delia, Sansotta and Donato19 had risk of detection bias, as presence of diarrhoea was assessed through participants’ diary and diarrhoea was not clearly defined. For similar reasons detection bias was showed by Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 and Giralt et al.Reference Giralt, Regadera and Verges21
Results are also inconsistent. Chitapanarux et al'sReference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 lower NNTB indicates higher efficacy of their treatment, but with a probiotic (Infloran) two orders of magnitude-less potent, containing less bacterial strains and administered less frequently than the probiotic (VSL#3) used by Delia et al.Reference Delia, Sansotta and Donato19 This could be because of the small sample size, reflected in the wider CI, by which the true efficacy could be near the CI higher boundary. Similarly, even if less strikingly, Germain et al'sReference Germain, Desjardins, Demers and Dagnault20 standard-dose group showed an effect size comparable with that by Delia et al.Reference Delia, Sansotta and Donato19 but with a probiotic two orders of magnitude-less potent. Moreover, Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 showed internal dose–effect inconsistency, as the standard-dose group had higher benefit (NNTB = 5·5) than the high-dose group (NNTB = 10·3).
Dose–effect inconsistency between Delia et al.Reference Delia, Sansotta and Donato19 and Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 could be due to a true effect. Delia et al.Reference Delia, Sansotta and Donato19 administered treatment only during radiotherapy, whereas Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 commenced treatment 7 days before starting radiotherapy, perhaps conferring increased prophylaxis. Moreover, Infloran capsule form might have superior gastric resistance compared with VSL#3 sachets, possibly decreasing potency discrepancy. Final formulation has an important effect on probiotic therapy efficacy.Reference Whorwell, Altringer and Morel35, Reference Del Piano, Carmagnola and Ballare36 In addition, Giralt et al.Reference Giralt, Regadera and Verges21 were the only investigators who found a non-beneficial effect of probiotics, albeit with no statistically significant results, and used a probiotic of one order of magnitude-less potent than the one used by Chitapanarux et al.Reference Chitapanarux, Chitapanarux, Traisathit, Kudumpee, Tharavichitkul and Lorvidhaya22 and Germain et al.Reference Germain, Desjardins, Demers and Dagnault20 Therefore, Giralt et al'sReference Giralt, Regadera and Verges21 results are consistent with a dose/effect argument.
Heterogeneity of effects is not surprising considering that probiotics include a plethora of microorganism species, with different behaviour in the intestine, and subject to diverse manufacturing methods. This leads to multiple treatment options, including strains used, potency, formulation and therapeutic regimen.
The probiotic preparations used in the included papers have been studied in connection with the treatment of other conditions. Infloran efficacy has been shown in paediatric settings for the treatment of acute diarrhoeaReference Lee, Lin, Hung and Wu37, Reference Rerksuppaphol and Rerksuppaphol38 and prevention of necrotising enterocolitis.Reference Lin, Su and Chen39 The efficacy of VSL#3 has been shown in the prevention of pouchitis,Reference Wall, Schirmer, Anliker and Tigges40 in the treatment maintenance of ulcerative colitisReference Heselmans, Reid, Akkermans, Savelkoul, Timmerman and Rombouts34, Reference Tursi, Brandimarte and Papa41 and in the reduction of diarrhoea in enterically fed patients.Reference Frohmader, Chaboyer, Robertson and Gowardman42 The British National Formulary lists VSL#3 for pouchitis prevention.43 Lactobacillus casei DN-114001 has showed efficacy in preventing or reducing the severity of infectious diarrhoea.Reference Hickson, D'Souza and Muthu44–Reference Turchet, Laurenzano, Auboiron and Antoine46 Bifilact has not received much attention in the literature.
Conclusions
When the above considerations are applied to the GRADE system of evidence quality rating,Reference Balshem, Helfand and Schunemann47 there is low-quality evidence that probiotics are beneficial in RID prevention: the large effects and their precision would suggest a high quality of the evidence, but this is downgraded by effect inconsistency and risk of detection bias (Table 6).
Table 6 Clinical evidence profile: probiotics for the prevention of radiotherapy-induced diarrhoea in adultsFootnote a
Notes: a Adapted from the modified GRADE profile25. A summary of findings is not included as the effect sizes of the individual studies (Table 5) were not combined with a meta-analysis.
b The largest study, accounting for 55% of the total number of patients, has serious risk of detection bias, as stool consistency was subjectively assessed and diarrhoea was not clearly defined.
c There is a dose/effect inconsistency between studies and within a study.
Abbreviation: RCT, randomised clinical trial.
Some of the probiotics investigated in the included studies have been shown to be effective in other gastrointestinal conditions, but this evidence cannot be immediately transferred to RID. Before advising towards the clinical use of probiotics in RID prevention, further trials, especially involving the promising VSL#3 and Infloran, are essential. Where appropriate, trials should describe randomisation, use intention to treat analysis, report effect sizes and assess diarrhoea objectively.
This review had limited scope and objectivity could not be improved without a second reviewer. Making sense of the disparate range of interventions in studies on probiotics is challenging. A meta-analysis on the prevention of RID with probiotics was intentionally avoided. Rather than numerically combining the disparate alternatives of this heterogeneous area of therapy, studies were considered in their own merit, to highlight meaningful results of individual studies and provide directions for future trials.
The studies did not assess participants’ perceived benefits, failing to provide further insight into probiotics efficacy. Moreover, patients with RID might develop chronic gastrointestinal symptoms, requiring long-term and often expensive treatment.Reference Andreyev4 Long-term follow-up of trial participants are needed to assess probiotic potential in reducing chronic complications and to provide insight of true probiotics clinical cost-effectiveness.
Acknowledgements
This review was self-funded as part of an MSc in Advanced Practice (Cancer Nursing) at King's College, London, UK.
