Efficacy of 5-Nitroimidazoles for the Treatment of Giardiasis: A Systematic Review of Randomized Controlled Trials

Background Giardiasis is one of the most common causes of diarrheal disease worldwide and 5-nitroimidazoles (5-NI) are the most commonly prescribed drugs for the treatment of giardiasis. We evaluated the efficacy of 5-nitroimidazoles (5-NI) in the treatment of giardiasis in a systematic review of randomized controlled trials (RCTs). Methodology/Principal Findings We conducted a comprehensive literature search in PubMed-Medline, Scopus, Web of Science and Cochrane Library for RCTs evaluating the efficacy of 5-NI vs. control (placebo or active treatment) on parasitological cure in patients with parasitologically-demonstrated giardiasis. The search was performed in May 2013 with no language restriction by two authors independently. The efficacy outcome was parasitological cure, and harmful outcomes were abdominal pain, bitter or metallic taste, and headache. We included 30 RCTs (n = 3,930). There was a significant and slightly higher response rate with 5-NI in giardiasis treatment (RR 1.06, 95%CI 1.02–1.11, p = 0.005). There was high heterogeneity among studies (I2 = 72%). The response rates for metronidazole, tinidazole and secnidazole were similar (RR 1.05, 95%CI 1.01–1.09, p = 0.01; RR 1.32 95%CI 1.10–1.59, p = 0.003; and RR 1.18 95%CI 0.93–1.449, p = 0.18, respectively). On subgroup analyses, the response rates did not vary substantially and high heterogeneity persisted (I2 = 57%–80%). Harmful outcomes were uncommon, and 5-NIs were associated with lower risk of abdominal pain, and higher risk of both bitter or metallic taste and headache. Conclusions Studies investigating the efficacy of 5-NI in giardiasis treatment are highly heterogeneous. 5-NIs have a slightly better efficacy and worse profile for mild harmful outcomes in the treatment of giardiasis in comparison to controls. Larger high quality RCTs are needed to further assess efficacy and safety profiles of 5-NI.


Introduction
Giardiasis is an intestinal illness caused by a flagellated protozoan parasite, Giardia lamblia (syn. G.intestinalis and G.duodenalis). The World Health Organization (WHO) estimates that 3 billion people reside in places with giardiasis prevalence of around 30%, and suggests that there are almost one billion cases of giardiasis, contributing to 2.5 million deaths annually from diarrheal disease [1]. In recent years, epidemiology of giardiasis in developed countries has been changing with increasing international travel and migration from highly endemic countries [2]. Approximately, 20,000 new cases of giardiasis are reported annually in the United States [3]. Due to its increasing global burden, and its developmental and socioeconomic impact on infected individuals, Giardia has been included in the 'Neglected Disease Initiative' of the WHO since 2004 [4].
The most common antibiotics used for the treatment of giardiasis are the 5-Nitroimidazoles (5-NIs); these include metronidazole, tinidazole, secnidazole and ornidazole, of which metronidazole is the most common [5]. Alternative agents which are less commonly used in giardiasis treatment are quinacrine, furazolidone, benzimidazoles (albendazole and mebendazole), paromomycin, bacitracin zinc, chloroquine and nitazoxanide [5]. Depending on local epidemiology, availability, and cost, these drugs have been widely available for the curative treatment of cases; however, several reports of treatment failure have been reported [6,7,8]. With the advent of newer agents which might have similar efficacies as 5-NIs,and also offer an added advantage of more simplified regimens, fewer adverse effects or less drug resistance, it is of considerable interest to determine whether 5-NIs are still the best available option in the treatment of giardiasis.
Three previous systematic reviews and/or meta-analyses [9,10,11] have evaluated efficacies of antigiardial drugs in the treatment of giardiasis. All three varied in study designs and study aims, but none compared efficacy of 5-NIs as a group in comparison to other antigiardial drugs. Against this background, we performed a systematic review of the literature to identify RCTs comparing the efficacies of 5-NIs with a control with the aim of assessing effectiveness of 5-NIs in the treatment of giardiasis. We hope to provide policymakers and practitioners with a convenient and evidence-based summary of the primary literature on which to base their decisions.

Data sources and searches
A comprehensive literature search using PubMed-Medline from database inception through May 13, 2013 The following predetermined inclusion criteria were used: (i) RCTs evaluating the efficacy of 5-NI in comparison with a control (placebo, active treatment); (ii) study population of patients with parasitologically-demonstrated giardiasis; (iii) study in any language. An active treatment group is a control group receiving comparator drug (5-NI or non-5-NI) for the treatment of giardiasis. Our exclusion criteria were: (i) no control group; (ii) efficacy data (parasitological cure rates) were not available or could not extracted for the study groups.

Study selection and data extraction
A list of retrieved articles was reviewed independently by 3 investigators (AVH, VP and AD) in order to choose potentially relevant articles, and disagreements about particular studies were discussed and resolved by consensus.
Two reviewers (VP and AD) independently extracted data from studies. The following information was extracted: age, gender, geographic location, study setting, diagnostic test for giardiasis, type of 5-NI and dose/duration, comparator drug and dose/ duration, follow up time. Extracted beneficial outcome was parasitological cure rate and harmful outcomes were abdominal pain, bitter or metallic taste, and headache. One other author (AVH) reviewed the extractions for inconsistencies, and the three investigators (AVH, VP and AD) reached consensus.

Evaluation of study quality
The quality of all included trials was assessed using a 5-item instrument developed and validated by Jadad [12]. The 5 items in this scale include i) description of randomization, ii) appropriateness of randomization, iii) description of blinding, iv) adequacy and appropriateness of blinding, and v) description of withdrawals and dropouts. Study quality was assessed independently by two investigators (VP and AVH). Disagreements were resolved by consensus. A score of 0-2 was considered as low quality trial and a score of 3-5 was considered high quality trial.

Data synthesis and analysis
Our systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Text S1, available as supporting data) [13]. A high degree of heterogeneity among studies was expected and therefore a formal meta-analysis was a secondary aim. Taking into account the sources of heterogeneity, several subgroup meta-analyses were prespecified: (i) type of 5-NI used, (ii) excluding studies with two types of 5-NI comparisons, (iii) study setting (outpatient vs hospitalized), (iv) Jadad score ($3 vs ,3), (v) type of main analysis (intention-totreat vs per-protocol), (vi) sample size (,100 vs $100 patients), (vii) ordered by year of publication. DerSimonian and Laird random effects models were used for meta-analyses [14]. We used the inverse variance (IV) or Mantel-Haenzel (MH) method to calculate pooled RRs and 95% CIs, depending on the absence or presence of scarce outcomes, respectively. When efficacy of two 5-NIs was compared, the 5-NI arm was the arm with the larger sample size. Statistical heterogeneity was evaluated with the Cochran x 2 and the I 2 statistics. I 2 values of 30-60% represented a moderate level of heterogeneity. A P value of ,0.1 for x 2 was defined as indicating the presence of heterogeneity. To examine bias in the results of the meta-analyses, the Egger's test was used to evaluate asymmetry of the funnel plots. Asymmetry of the funnel plots should not be equated with publication bias, as asymmetry can be caused by true heterogeneity among study results, poor

Author Summary
Giardiasis is a major diarrheal disease with worldwide distribution. 5-nitroimidazoles, which include metronidazole and tinidazole, are the most commonly used drugs in the treatment of giardiasis. In recent years, many other drugs with variable efficacies and adverse effects have been proposed for the treatment of giardiasis. No systematic review has evaluated efficacy of 5-nitroimidazoles as a group in comparison to the other antigiardial drugs. In this context, we performed a systematic review of the literature to identify randomized controlled trials comparing the efficacies of 5-nitroimidazoles with a control drug with the aim of assessing effectiveness of 5nitroimidazoles in the treatment of giardiasis. Four research databases were searched; 30 trials with 3,930 subjects met our inclusion criteria. Results show that there was a high variation of study outcomes between included studies. 5-nitroimidazoles were associated with higher giardiasis cure rates than controls; also, 5-nitroimidazoles are associated with lower risk of abdominal pain, and higher risks of bitter or metallic taste and headache than controls. 5-Nitroimidazoles for the Treatment of Giardiasis methodological quality, reporting biases, and chance. We used Review Manager (RevMan 5.0, Oxford, UK; The Cochrane Collaboration, 2008).

Eligible studies
Our search identified 333 publications ( Figure 1). After removing duplicates and screening titles of the studies, 75 articles were selected based on relevance to the study topic. After screening the abstracts of these potentially relevant articles, 42 were selected for full-text review based on relevance to the study topic ( Figure 1). Thirty randomized controlled trials (n = 3,930) in twenty-nine studies that reported efficacy data of 5-NIs in comparison to a control were included in the systematic review and meta-analyses. The reasons for exclusion of the remaining 13 articles are listed in Figure 1. Table 1 summarizes the main characteristics of the included studies. All trials were conducted in countries endemic to giardiasis. Of the 30 trials included, 22 were in outpatient population; 7 in hospitalized population; one trial did not report the study setting. All patients included in the trials had parasitologically-demonstrated giardiasis. Twenty-four trials were in pediatric population (,18 yrs); 4 trials in both adult and pediatric population; two trials in patients of age .17 yrs. The post-treatment follow-up time varied from 3 days to 5 weeks. A total of 3,930 patients were included in the meta-analysis with sample size ranging from 23 to 502; all but one of the trials included active treatment controls (Table 1). NA = not available; * = Mean (range); # = median (range); J = children got half the adult dose.

Study quality assessment and publication bias
Using the Jadad scale, 13 trials were identified as high quality (Table S1, available as supporting data). All studies were described as randomized, 13 of them appropriately described the generation of the sequence of randomization, but none of studies were appropriately blinded. Twenty-nine studies appropriately described withdrawal and dropouts. Publication bias assessed by funnel plot showed some asymmetry around the point estimate, especially for small sample size studies indicating presence of bias ( Figure 2). The Egger test did not suggest asymmetry of the funnel plot (p = 0.1).  Efficacy of 5-NIs in the treatment of giardiasis and metaanalyses of subgroups of studies We found a significant and slightly higher cure rates (RR 1.06, 95%CI 1.02-1.11, p = 0.005) (Figure 3). There was high heterogeneity among studies (I 2 = 72%). When stratified by type of drug, efficacy of 5-NIs did not vary significantly and high heterogeneity persisted (I 2 = 57%-80%): metronidazole (RR 1.05, 95%CI 1.01-1.09, p = 0.01); tinidazole (RR 1.32 95%CI 1.10-1.59 p = 0.003); and secnidazole (RR 1.18 95%CI 0.93-1.49 p = 0.18) (Figure 4). There was no study comparing the drug ornidazole to a control group.

Discussion
We found that RCTs evaluating the efficacy of 5-NIs for giardiasis treatment are highly heterogeneous in terms of study design and outcomes. Heterogeneity could not be diminished after performing several pre-specified subgroup analyses. The quality of RCTs is mostly low, especially because of lack of double blinding. 5-NIs are associated with a slightly higher giardiasis cure rates than controls; also, 5-NIs are associated with lower risk of abdominal pain, and higher risks of bitter or metallic taste and headache than controls.
Since the first publication of metronidazole in the treatment of giardiasis by Schneider et al [15] more than 50 years ago, 5-NI compounds (mainly metronidazole and tinidazole) have become an important component of the antigiardial armamentarium in many parts of the world, owing to their efficacy, relative safety, universal availability, and cost-effectiveness. However, during these last five decades, 5-NI compounds, principally metronidazole, have also been usually prescribed for several other indications, including gingivitis, bacterial vaginitis, part of the combination treatment for H. pylori, infections with Clostridium difficile and other anaerobic bacteria, amebiasis, and as prophylaxis in colorectal surgery [16,17].This wide spectrum usage of 5-NIs could have led to an increased occurrence of G. lamblia resistance; in fact, Giardia resistance towards common antigiardials has been demonstrated or induced in vitro [18] and also cross-resistance between metronidazole and tinidazole has been demonstrated [19]. Most of the therapeutically used antigiardial drugs, including metronidazole cause severe side effects and are not well tolerated by many patients and clinical resistance to medication has been observed for all common drugs in up to 20% of giardiasis cases. Treatment failure may be due to both host factors (e.g. low patient compliance due to side effects) and parasite resistance [5,20]. The present study systematically reviewed all available data from trials examining parasitological outcomes, and comparing efficacy of different 5-NI drugs, doses and regimens. The results obtained, suggest that 5-NI continue to be efficacious for giardiasis. However, it should be stated that we did not find any trial comparing ornidazole to a control group, and so we cannot make recommendations for this drug. Three previous systematic reviews on the treatment of giardiasis have been published (Table S2). In a systematic review by Zaat and colleagues [10], 31 RCTs published up to 1997 were included (n = 2,988). Three databases were searched, pseudo-randomized trials were included and no language restriction was used. Any trial of treatment of giardiasis comparing drugs or treatment regimens with placebo or other drugs/regimens were included. Metronidazole was found to be equally effective in parasitological cure as other longer therapies such as furazolidone. Tinidazole seems to be the most effective single-dose therapy in terms of parasitological cure compared to other short therapies, having, at the same time, relatively fewer harmful effects such as diarrhea at the end of follow-up. Trials were also found to be clinically and statistically heterogenous. Majority of the included studies were low on methodological quality as per Cochrane Collaboration guidelines. A recent meta-analysis by Solaymani-Mohammadi et al [11] included 8 RCTs (n = 900) published up to 2010 comparing effectiveness and safety of metronidazole vs. albendazole for the treatment of giardiasis. Six databases were searched, and no language restriction was applied. Effectiveness of albendazole was found to be comparable to metronidazole. Patients treated with albendazole also tended to have fewer side effects such as metallic taste and anorexia. Included trials were found to have moderate heterogeneity of effects and the quality was low in 7 of them (Jadad score 0-2).
A recent systematic review by Granados et al [9] included 19 studies (n = 1,817) published up to 2012 and evaluated the relative effectiveness of alternative antibiotic regimens for treating adults or children with symptomatic giardiasis. Six databases were searched, and no language restriction was used. All RCTs comparing metronidazole administered for five to 10 days with any of the following drugs: metronidazole (single dose), tinidazole, albendazole, mebendazole, and nitazoxanide were included. The primary outcomes were parasitological and clinical cure. They concluded that albendazole may be of similar effectiveness to metronidazole, may have fewer gastrointestinal and neurological side effects, and has the advantage of a simplified regimen; evidence was considered to be of moderate quality based on GRADE methodology. Included trials had moderate heterogeneity of effects.
Our study is the first to examine the efficacy of 5-NIs as a group in comparison to other antigiardial drugs in the treatment of giardiasis. Search criteria in our review were not restricted by language thereby avoiding 'tower of babel bias' [21]. Several RCTs included in our meta-analysis were deficient in quality (i.e. high risk of bias) in included trials. Given that most of studies did not appropriately use blinding, the probability of information bias and inflation of efficacy of 5-NIs is present. Also, since heterogeneity among studies was expected, we had pre-specified an extensive list of specific study variables for subgroup analyses ( Figure S1-S5). On subgroup analysis both low quality studies and high quality studies (by Jadad scoring) showed higher efficacy of 5-NIs vs. controls, while only results from high quality studies achieved significance. Effect sizes remained fairly constant on the rest of subgroup analyses suggesting that heterogeneity of included trials might not have adversely affected our results. In our metaanalysis 28 trials included pediatric population and 6 trials included adult population. Including subjects of different age groups as well as with varied clinical manifestations and a variety of clinical settings allows us to generalize our findings of effectiveness of 5-NI treatment to all age groups and all kinds of symptomatic giardiasis. 5-NIs were associated with higher risk of bitter or metallic taste and headache. Though, these would be considered as minor adverse effects, drug tolerability and adverse events could potentially impact patient compliance. Our findings suggest that 5-NIs have high cure rates and reasonable safety profiles and in absence of better alternative drugs remain the drug the choice in the treatment of giardiasis.  Our study has specific limitations that need to be considered in the interpretation of our findings. There was heterogeneity in some of the relevant study design aspects (time of follow-up, different doses of the drugs, heterogeneity of participants, number of parasitological exams at follow-up, and laboratory techniques across the studies) which made some results difficult to interpret and precluded us from a more confident and robust conclusions about benefits and harms associated with the use of 5-NIs in the treatment of giardiasis. The Egger's test did not suggest asymmetry of the forest plot, but some degree of publication bias can be expected. Results should also be interpreted with caution in light of high proportion of low quality RCTs. Also, we used one of the several tools to evaluate risk of bias, the Jadad score. Unfortunately, this tool does not evaluate other potential and important sources of bias such as concealment of the randomized allocation and selective reporting of outcomes. Despite these limitations, this present review represents the most up-to-date, comprehensive, and systematic attempt to assess the efficacies of 5-NI drugs as in group in the treatment of giardiasis.
Studies investigating the efficacy of 5-NIs in the treatment of giardiasis are highly heterogenous. Though information available from RCTs on the use of 5-NI allow us to confirm that these drugs are still a good option for the treatment of giardiasis, there is a need for well designed, high quality RCTs to further assess safety and efficacy profiles of 5-NIs.