Is It Useful To Treat Blastocystis sp.? A double-blind placebo-controlled randomised trial

The primary objective of this study was to evaluate the usefulness of metronidazole in patients with gastrointestinal symptoms harbouring only Blastocystis sp. In addition, we explored whether Blastocystis subtype or concomitant parasitic infection detected by polymerase chain reaction (PCR) may inuence treatment outcome.


Abstract Background
Blastocystis sp. is a protist with a worldwide distribution and able to colonise the gut of humans and of a great variety of animals. It is unclear whether it is just a commensal of a healthy gut microbiota or an infectious parasite that needs to be eradicated. Currently no treatment has proven its usefulness for patients complaining of gastro-intestinal symptoms and found to have Blastocystis sp.
The primary objective of this study was to evaluate the usefulness of metronidazole in patients with gastrointestinal symptoms harbouring only Blastocystis sp. In addition, we explored whether Blastocystis subtype or concomitant parasitic infection detected by polymerase chain reaction (PCR) may in uence treatment outcome.

Methods
Adults with persistent gastrointestinal symptoms (> 14 days) visiting a primary care physician and in whom stool microscopy revealed only Blastocystis sp. were included. Eligible patients were randomised to receive ten days of metronidazole or placebo, followed by a crossover if still symptomatic. Stool samples were tested for 11 other protozoa with an in-house PCR and Blastocystis subtypes were determined by PCR and sequencing.
After the crossover, again no differences in improvement of symptoms were seen between groups (placebo: 53% (8/15); metronidazole: 50% (8/16)). The in-house PCR was positive for other protozoa in 25% (10/40) of the patients. The protozoa identi ed were Dientamoeba fragilis (5), Entamoeba dispar (3) and Cyclospora cayetanensis (2). The most frequent Blastocystis subtypes were ST4 (11/36) and ST2 (10/36 Blastocystis sp. is a protistan living in the intestinal tract of humans and a wide range of animals, such as birds, reptiles, amphibians, sh, insects. It has high genetic diversity, with 22 different subtypes [1]. Of the ten subtypes infecting humans, subtype 3 (ST-3) is the single most frequently found. [2] Blastocystis sp. has a worldwide distribution with higher prevalence in low-and middle-income countries, which is probably due to poor hygienic conditions and exposure to animals, as well as consumption of contaminated food and water. [3] Travellers to tropical countries seem at increased risk of acquiring Blastocystis sp. [4][5] Asymptomatic individuals from temperate, high-income countries can also carry it, prevalence ranges from 14% in Spain [6] to 56% in Ireland [7] Blastocystis sp. pathogenicity has long been a matter of debate and it has been postulated that subtypes may differ in pathogenicity and sensitivity to drugs. [8] The usefulness of treating Blastocystis sp. infection is unclear. Several observational studies have shown no association between the presence of Blastocystis sp. in the digestive tract and symptoms. [9][10][11][12] Similarly, in a case series of 100 patients with stool microscopy positive for Blastocystis sp. in Taiwan, only 10 had gastrointestinal symptoms and all improved without receiving any speci c therapy. [13] A randomised, double-blind, placebo-controlled trial by Heyland et al. on a paediatric population compared the effect of cotrimoxazole on recurrent abdominal pain and no difference in pain reduction was seen between the 2 groups. [14] Lastly, various in vitro studies report evidence of Blastocystis drug resistance. [15][16][17] Some studies suggest that treating Blastocystis sp. may improve non-speci c gastrointestinal symptoms, such as nausea, anorexia, abdominal pain, bloating, atulence, as well as acute and chronic diarrhoea. subjects and enrolled 76 patients complaining of diarrhoea with Blastocystis sp. as the only identi ed microorganism and without other identi able causes. Among subjects receiving metronidazole, 88% reported cessation of symptoms one month after therapy compared to 14% in those receiving placebo. [24] Our study main objective was to evaluate if, in a primary care setting, treatment with metronidazole would improve persistent gastrointestinal symptoms of patients without signi cant comorbidities found to harbour Blastocystis sp. by stool microscopy. The secondary objective was to describe evolution of gastro-intestinal symptoms with treatment. A post-hoc analysis was performed to determine Blastocystis subtypes and to test stools with an in-house real-time PCR for protozoa other than Blastocystis sp.

Methods
In this double-blind, randomised crossover trial, all physicians of the general primary care outpatient clinic of the University Hospital of Lausanne, practitioners of the Tropical and Humanitarian division of the Geneva University Hospitals and ten private general practitioners of the neighbouring region acted as recruiters. They evaluated their patients according to their usual practice. When a parasitological exam was deemed necessary, they sent three stool samples xed in sodium acetate-acetic acid formalin solution (SAF) for parasitology microscopy and for detection of stool antigen of Giardia lamblia. Investigators received a noti cation for specimens positive for Blastocystis sp. and contacted the patients for inclusion in the study.
The inclusion criteria were age ≥ 18 years, gastrointestinal symptoms for more than 14 days, a collection of 3 separate stool specimens, identi cation of Blastocystis sp. as the only pathogen on microscopy and negative Giardia lamblia stool antigen. Exclusion criteria were as follows: signi cant alteration of general condition, axillary temperature > 37,5 °C, visible blood in the stool, known chronic digestive symptoms, immunosuppression (due to drugs or disease), oncological disease, pregnancy or lactation, weight loss > 10% from the usual weight, antimicrobial treatment within the last three months. Patients being treated with anti-vitamin K or disul ram were also excluded as metronidazole is known to increase their activity.
After signing the informed consent, patients had to complete a questionnaire evaluating the following symptoms: number of stools per day during the three previous days, presence of soft or unformed stool, as well as the severity of abdominal pain, bloating and atulence on a scale from 0 to 10. After that, patients were randomised with a 1:1 ratio in a double-blind fashion to a regimen of metronidazole 500 mg or placebo three times a day for ten days.
The University Hospital Pharmacy oversaw the capsules production and drug packages preparation. Metronidazole and placebo capsules were visually indistinguishable. Drug packages were prepared with four, visually indistinguishable containers labelled A to D, and lled with either metronidazole or placebo.
The Pharmacy generated a randomisation list with the containers' content to ensure that each patient received sequentially the two drugs. The randomisation list was kept in a secure place within the Pharmacy and was not available to the physicians and the investigators.
Treating physicians gave a drug package to the included patients according to the administration sequence list. Ten to fourteen days after the end of treatment, we assessed the symptoms again with the questionnaire. In addition, patients had to self-evaluate the impact of the treatment on their general wellbeing according to an ordinal scale consisting of four categories: no improvement -slight improvement -signi cant improvement -cured.
The second drug was given to patients in both groups who considered themselves as not cured.
Symptoms were again assessed 10 to 14 days after the end of the 2nd-course treatment using the questionnaire as mentioned above.
The primary study endpoint was the proportion of participants reporting any improvement (de ned as either "slight improvement", "signi cant improvement" or "cured") of their general wellbeing. Secondary endpoints were the proportion of participants reporting unformed or soft stool, the mean number of stools per day, maximal abdominal pain, bloating and atulence during the three previous days.
After the end of the trial, we determined the Blastocystis subtype by sequencing the 18SrRNA gene and tested 40 frozen stool samples for 11 protozoa (Entamoeba dispar, Entamoeba histolytica, Cryptosporidium hominis, Cryptosporidium parvum, Encephalitozoon spp. Entamoeba moshkovskii, Entamoeba polecki, Dientamoba fragilis, Giardia lamblia, Cyclospora cayetanensis, Cystoisospora belli) by an in-house TaqMan probe-based real-time PCR developed at the Swiss Tropical and Public Health Institute (SwissTPH) (See Appendix A). Samples positive for Blastocystis 18SrRNA real-time PCR were subsequently ampli ed by PCR to obtain a 600 bp long fragment suitable for sequencing and determination of the Blastocystis subtype according to Scicluna et al. [25]. All details of the molecular methods are provided in the supplementary material. (Molecular Microbiology Procedures) We calculated the sample size based on the proportion of patients still reporting unformed or soft stools after treatment in the study from Nigro et al. [24] With a two-sided 5% signi cance level, a power of 90% and anticipating a dropout rate of 20% we estimated that we would need 200 patients to show a reduction of the proportion of patients reporting unformed or soft stool from an initial 50-25%. Statistical analysis included a chi-square analysis for dichotomous outcomes and a Wilcoxon rank-sum test for ordinal outcomes. We chose an alpha level of .05 for all statistical tests. Statistical analyses were performed with STATA 15. All methods were carried out in accordance with relevant guidelines and regulations. The study was registered on ClinicalTrials.gov under the number NTC01521403 on the 6th of November 2012.

Results
From December 2012 to April 2017, we screened 474 patients of which 424 refused or did not met inclusion criteria (Fig. 1) Thus, fty patients met the eligibility requirements and were enrolled. Twenty-ve patients were randomised to each group with comparable baseline characteristics. (Table 1)  Missing data = 1 6 (3-8) Missing data = 1 The mean age was 41.9 years, 82% reported travel abroad in the previous 12 months. The most frequent destinations were the Indian subcontinent and Southeast Asia, 78% reported loose stools. Median (interquartile range) number of bowel movements, maximum abdominal pain, bloating and atulence severity were 2 (1-3), 4 (0-6), 5 (1-7), 6 (3-8) respectively (Table 1). Before the start of the study, eight patients withdrew their consent or were lost to follow-up (Speci c reasons mentioned in the study ow, Fig. 2).
Patients in both treatment groups experienced an improvement in symptoms 10 to 14 days after treatment completion. In the intention-to-treat analysis, 48% of patients in the metronidazole group reported an improvement of their general wellbeing compared to 44% in the placebo group. This difference was not signi cant (χ²(1, n = 50) = 0.08, p = .78, OR = 1.17, 95% CI 0.062-2.29) and the effect size was very small (Cohen's h = 0.08). The proportion of patients reporting abnormal stool consistency decreased from 80-44% in the placebo group and from 76-36% in the metronidazole group. Median abdominal pain (interquartile range) decreased from 4 (2-6) to 1 (0-5) in the placebo group and from 3 (0-6) to 1.5 (0-3.5) in the metronidazole group (Table 2).  In the 2nd treatment phase, 64% (16/25) from the placebo group received metronidazole, and 60% (15/25) from the metronidazole group received a placebo. Again, no differences in improvement of symptoms were seen between treatment groups (53% (8/15) in the placebo group versus 50% (8/16) in the metronidazole group) ( Table 3). No serious adverse events or unintended events were registered.  For the post-hoc study aimed at determining the presence of protozoa by PCR and Blastocystis sp. subtyping, enough samples were available for 80% (40/50) of patients. Twenty-ve per cent (10/40) had a positive PCR for other protozoa undetected by microscopy (Fig. 3), distribution did not differ between the 2 groups.
In the placebo group the following protozoa were identi ed by PCR: Dientamoeba fragilis = 3, Entamoeba dispar = 2, Cyclospora cayetanensis = 1. In the metronidazole group the PCR showed the presence of the following additional protozoa: Dientamoeba fragilis = 2, Entamoeba dispar = 1, Cyclospora cayetanensis = 1. Blastocystis PCR was negative in 4/40 stool specimens positive for Blastocystis sp. by microscopy. Blastocystis subtyping was successful in 31/36 patients. The most frequent subtypes were ST4 (11) and ST2 (10), followed by ST3 (8) and ST1 (2). Analyses strati ed by subtype or the simultaneous presence of other protozoa did not show any difference in the bene t of treatment with metronidazole.

Discussion
In this double-blind placebo-controlled randomised study, there was no difference in improvement in general wellbeing and speci c digestive symptoms in patients infected with Blastocystis sp. and treated with metronidazole or placebo.
A strength of our study resided in the stringent inclusion and exclusion criteria, which allowed to test the usefulness of metronidazole on a very homogenous population. Most previous studies included more heterogeneous populations, such as patients with comorbidities. However, at the same time, our approach limited patient recruitment which ultimately prevented us from reaching the planned sample size of 200 patients. Indeed, we decided to stop the study after 52 months due to slow recruitment.
It can be argued that viral or bacterial agents causing diarrhoea have not been investigated prior to inclusion, but such infectious agents do usually not cause abdominal symptoms lasting more than 14 days. Patients who had received antibiotics in the previous three months were excluded to avoid possible delayed effect by a previous anti-infectious treatment. We also decided purposefully not to perform stool tests after the two treatment periods, because in general practice only clinical cure is relevant. To show a microbiological effect of metronidazole was not the aim of our pragmatic study.
Three out of the 4 randomised trials which investigated the e cacy of antiparasitic drugs for symptomatic Blastocystis sp. infections were either focusing on a different drug than metronidazole or on paediatric patients. It is therefore di cult to compare our results to those studies [14,[22][23][24].
The recommendation for metronidazole use in adult patients with gastrointestinal symptoms and Blastocystis sp. infection is mainly based on a study from Italy. [24] Several methodological reasons may explain the divergent results between the Italian study and ours. First, in the Italian study physicians who collected the data were also aware of patient group allocation. Second, the authors assessed abdominal pain, fatigue and stomach gurgling without clearly de ning measurement criteria. Finally, the mean duration of diarrhoea was 10 ± 3 days for the metronidazole group and 11 ± 2 days in the placebo group. Therefore, one could not reasonably exclude the presence of further undetected pathogen sensitive to metronidazole. In contrast, our study was double-blinded, we assessed symptoms with a linear analogue scale and, although we did not measure the duration of symptoms, we included only patients complaining of persistent symptoms, i.e. of at least 14 days. We may point out that including only patients with persistent symptoms may have led to the selection of patients with travel related postinfectious irritable bowel syndrome for which anti-parasitic drugs are ineffective.
An in-house PCR panel for intestinal protozoa found a concomitant pathogenic protozoa in 25% of our stool specimens. Our results therefore support to some extent the widely believed theory that the use of metronidazole has sometimes a positive effect on patients with Blastocystis infection, because the drug is active against a concomitant infection with another protozoa. [26] In 5 subjects we identi ed Dientamoeba fragilis, a protozoon commonly treated by metronidazole. However, this was not the case, but the numbers were too small to draw any rm conclusion. It is useful to remember that the diagnosis of Dientamoeba fragilis and Cyclospora cayetanensis by microscopy requires special stains while other protozoa such as Giardia lamblia are reliably detected.
The relative distribution of Blastocystis subtypes was different from previous studies. [27] We found ST4 (31%) and ST2 (28%) being the most frequent subtypes, and ST3 was present only in 22% of samples. These results suggest that a large proportion of our subjects acquired their Blastocystis sp. infection locally, as ST4 is rarely reported outside of Europe. [28] This nding is in line with other studies supporting the hypothesis that ST4 may have higher virulence. [29][30][31]