Traditional medicinal plants in the treatment of gastrointestinal parasites in humans: A systematic review and meta‐analysis of clinical and experimental evidence

Gastrointestinal (GI) parasites cause significant morbidity and mortality worldwide. The use of conventional antiparasitic drugs is often inhibited due to limited availability, side effects or parasite resistance. Medicinal plants can be used as alternatives or adjuncts to current antiparasitic therapies. This systematic review and meta‐analysis aimed to critically synthesise the literature on the efficacy of different plants and plant compounds against common human GI parasites and their toxicity profiles. Searches were conducted from inception to September 2021. Of 5393 screened articles, 162 were included in the qualitative synthesis (159 experimental studies and three randomised control trials [RCTs]), and three articles were included in meta‐analyses. A total of 507 plant species belonging to 126 families were tested against different parasites, and most of these (78.4%) evaluated antiparasitic efficacy in vitro. A total of 91 plant species and 34 compounds were reported as having significant in vitro efficacy against parasites. Only a few plants (n = 57) were evaluated for their toxicity before testing their antiparasitic effects. The meta‐analyses revealed strong evidence of the effectiveness of Lepidium virginicum L. against Entamoeba histolytica with a pooled mean IC50 of 198.63 μg/mL (95% CI 155.54–241.72). We present summary tables and various recommendations to direct future research.

Around 50 million new cases and 55,000 deaths caused by E. histolytica are reported annually (Cui et al., 2019;Lozano et al., 2012).
G. intestinalis is the third most common agent of diarrheal disease in children under 5 years of age globally (Lanata et al., 2013), with more than 280 million cases annually (Fink et al., 2020). Cryptosporidium parvum and C. hominis are the most frequently detected species of Cryptosporidium in humans and are responsible for approximately one million deaths yearly (Villanueva, 2017). Globally, Cryptosporidium spp. is estimated to be accountable for 30%-50% of the deaths in children under 5 years of age (Ochoa et al., 2004;Snelling et al., 2007).
GI parasitic diseases are associated with chronic and insidious effects on the health and nutritional status of those affected (WHO, 2012). Intestinal helminths negatively affect physical and mental development and reduce productivity and earning potential in adults (Guyatt, 2000). Protozoan parasites such as E. histolytica, C. parvum and G. duodenalis are often associated with diarrhoea and adversely affect growth and nutritional status, especially during infancy (Guerrant et al., 1999;Sullivan et al., 1991).
These GI parasitic infections continue to have a substantial socioeconomic impact, despite strong public health programmes, improved sanitation and mass drug administration (Webster et al., 2014). The most serious barrier to controlling many of these parasite species is partial or complete resistance in parasites to currently available synthetic drugs (Albonico et al., 2004;Picot et al., 2022). Moreover, drugs used to control parasites are usually expensive and unaffordable to low-income populations where infectious diseases are common (Nixon et al., 2020). Recurrence of infections is also common due to the lack of effective drugs, poor hygiene facilities and heavy environmental contamination with infectious parasite stages (Adam et al., 2016;Brooker et al., 2006).
To overcome these limitations and achieve more effective parasite management, discovering alternative sources for antiparasitic drugs is essential. Exploring potential botanical antiparasitics, typically inexpensive and abundant, appears to be a promising alternative in this context. Traditional botanical therapies have been used worldwide for many years to manage different GI parasitic diseases, for example, by practitioners of the Indian traditional medical system (Ayurveda) and Chinese traditional medicine. However, many indigenous plants with antiparasitic potential lack systematic scientific evaluation for their efficacy, mode of action and active chemicals. Out of 300,000 plant species, only 6% have been systematically investigated for their pharmacological effects, and only 15% have been evaluated phytochemically (Athanasiadou et al., 2007;Balandrin et al., 1993;Raskin et al., 2002).
In addition, plant products may sometimes induce cytotoxic effects due to the presence of multiple bioactive compounds. In many countries, there is no widespread regulatory system ensuring the safety of plant products, and they have not been adequately investigated toxicologically or analytically (Valerio & Gonzales, 2005).
With these gaps in mind, this comprehensive systematic review aimed to gather and synthesise the available literature on the efficacy of different plants and plant compounds against common human GI parasites and the toxicity profile of plants with antiparasitic activity.
Our goal was to provide up-to-date knowledge about the antiparasitic potential of traditional medicinal plants and highlight existing knowledge gaps to help direct future research.

| Study design
This systematic review with meta-analyses was designed and conducted according to the guidelines of Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) (Page et al., 2021

| Exclusion criteria
Review studies, case reports, letters to editor, short communications, commentaries and conference proceedings were excluded from the study. Studies in which plant medications were not the primary intervention were also excluded.

| Search strategy
The studies were identified using a comprehensive literature search of the databases PubMed, Scopus, Ovid Medline, Cochrane Library, CINAHL and the Indian Journals from inception to September 2021.
The reference lists of included experimental studies and excluded review articles were hand-searched to identify other potentially suitable studies.

| Study evaluation and selection
Reports sought for retrieval (n = 258) Reports unable to be retrieved (n = 53)

Reports (Full texts) assessed for eligibility (n = 205)
Reports excluded: n = 43 Study type not relevant (n = 21) Parasite not relevant (n = 13) No plant tested (n = 4) AnƟparasiƟc effect Not tested (n = 2) Not in English (n = 1) Parasite not described (n = 1) Results duplicated in another year (n = 1) Studies included in the quanƟtaƟve synthesis (n = 3) F I G U R E 1 PRISMA flow diagram. independent reviewers (SR and SA or AB) screened the titles and abstracts of search results for inclusion using Rayyan software (http:// rayyan.ai/reviews). Full-text articles were retrieved based on reviewer agreement and were reviewed independently by two reviewers (SR and SA or AB) for inclusion. Disagreements at each stage were resolved by discussion or consultation with a third review author (SA or AB) when necessary. Corresponding authors were emailed once to request a full-text copy if the full text could not be retrieved.

| Data management and extraction
Data were independently extracted from the full text of the included studies by two authors (SR and AZ) using data collection forms developed by the consensus of two authors based on the needs of the review. Data collection forms were pilot tested on four studies before full data extraction. If necessary, disagreements were resolved through consensus or consultation with a third review author.
The data extraction form included the following: (i) general characteristics (name of the first author, published year, country); (ii) participant characteristics (in vitro studies: cell culture; in vivo studies: details of animal models; RCTs: age, sex, severity); (iii) information on study design (design, sample size, randomisation, blinding method, diagnostic criteria, toxicity screening, statistical methods); (iv) details of interventions (plant species and family name, part used, dosage, administration route, frequency, duration, follow-up, parasite tested); (v) details of comparison interventions (type, administration route, frequency, duration); and (vi) outcome measures (antiparasitic effects).

| Risk of bias and quality assessment
The modified version of the Cochrane risk-of-bias tool 2 for random- The RoB 2 tool includes the following categories: random sequence generation, allocation concealment, blinding of participants, blinding of outcome assessors, completeness of outcome data, selective outcome reporting and other biases. The assessments for each item were rated as low risk, unclear risk or high risk.
The OHAT Risk of Bias Rating Tool categories are exposure or sequence generation, allocation concealment, identical housing, blinding research personnel, incomplete outcome data, confidence in the exposure characterisation, outcome assessment, selective outcome reporting and other biases. The evaluations for each item were graded as definitely low risk, probably low risk, probably high risk and definitely high risk.

| Data synthesis and meta-analysis
The body of evidence was synthesised qualitatively, and, where appropriate, meta-analyses were performed. Meta-analyses were conducted if at least three studies assessed the same plant against a particular parasite using the same measure of antiparasitic effect. Metaanalyses were conducted to investigate the effect of the plant against the parasite in question, and if a suitable comparator intervention (e.g. control drug) was used in at least three studies, an additional meta-analysis was performed to compare the mean difference of the effect.
The meta-analyses pooled the antiparasitic effects of the plant extract (e.g. half-maximal inhibitory concentration/IC 50 ) and their standard errors (SE) using a random effects model, which allows for differences in the treatment effect among studies. In studies where the SE was not reported, it was calculated from standard deviation For mean difference meta-analyses, the effect of each plant extract relative to a comparator was calculated using the difference in means between the study groups. The SE of mean difference (MD) was calculated using the following formula [SE (MD) = SD diff/ √N] (Higgins & Green, 2011). If a single study had more than one testing parameter eligible for inclusion in the meta-analysis, to account for the lack of independence with multiple study outcomes, a combined mean change and variance were first calculated using a fixedeffect model (Borenstein et al., 2021), which was then included in the mean difference meta-analysis using a random effects model. Metaanalyses were performed with JASP Team Version 0.16.3 software (Amsterdam, Netherlands).
Some studies measuring IC 50 did not report the effect size (particularly when the plant was considered inactive). In these instances, we imputed the values for missing effect sizes in two ways: (i) using the highest IC 50 value (the least effective) from the same study and (ii) using a previously established IC 50 value for inactive plants for the parasite (Nezaratizade et al., 2021). We then performed a sensitivity analysis to compare the meta-analysis results using data imputed using each of the two listed methods.

| Search results
The electronic database search retrieved a total of 7775 published articles. After removing duplicates, 5393 articles were assessed based on the title and abstract screen. Of these, 5135 records were excluded, and the rest, 258 articles, were sought for full-text retrieval.
Only 205 full texts could be retrieved and evaluated for inclusion.
After screening the full-text articles, 43 were excluded, and the rest, 162 articles, were included in the study (Figure 1).
In total, 507 plant species belonging to 126 families were tested against different parasites (see Tables S1-S6 for  Rubiaceae (n = 13).

| Parasites targeted
The protozoan parasites E. histolytica (n = 37) and G. duodenalis (n = 25) were targeted in most in vitro studies. This was followed by the use of non-parasitic models Pheretima posthuma and Caenorhabditis elegans (n = 32 each) (Figure 2). Hymenolepis nana, also known as Vampirolepis nana (n = 37) and Hymenolepis diminuta (n = 37), were the most evaluated parasites in vivo ( Figure 2). Swiss albino mice (n = 13) and Wistar rats (n = 10) were the most widely used in vivo experimental models. In most RCTs (n = 2), mixed infection of GI helminths was identified in the participants.

| Evaluation methods for plant antiparasitic activities
In vitro antiprotozoal activities were assessed by growth inhibition of trophozoites (n = 30) or trophozoite viability/killing activity (n = 15).
There were some dose limits of Butea frondosa and Embelia ribes/E. robusta and mixtures of B. frondosa, E. ribes and Mallotus philippinensis had no effect on the RCTs (Table S8).
The plant species (n = 5) belonging to the Artemisia genus were The methanol extract of the aerial parts of Ocimum basilicum showed significant activity on multiple parasites, including E. histolytica, G. lamblia, P. posthuma, E. eugeniae (earthworm) in vitro (Basha et al., 2011;Calzada et al., 2006;Osei Akoto et al., 2020). By contrast, the methanol extract of its leaves had no effect on E. histolytica in a different study (Quintanilla-Licea et al., 2014). Furthermore, dichloromethane/methanol (1:1) extract from the aerial parts of the same plant had no effect against E. histolytica and G. lamblia (Camacho-Corona et al., 2015). Different parts (twig, root bark, fruit and wood) and different extracts (water and organic) of Ziziphus mucronate had significant activity against C. elegans (Waterman et al., 2010). However, the water and organic extracts of its leaf and bark did not affect this non-parasitic model (McGaw et al., 2007).
Justicia spp. showed significant in vitro anthelmintic activity against the non-parasitic models P. posthuma and C. elegans (Ikbal F I G U R E 2 Parasites, parasitic or non-parasitic models tested using in vitro, in vivo or randomised control trials. Joshi et al., 2020). However, the same plants did not affect the protozoan parasite E. histolytica (Heinrich et al., 1992;Tona et al., 1999). By contrast, Lippia spp. exhibited in vitro efficacy against both helminths and protozoans, including G. intestinalis, E. histolytica and C. elegans (Calzada et al., 2006;Vera-Ku et al., 2010).

| Evaluation of toxicity
Testing plants for their toxicity was not an objective of many studies. Only 24 studies on 38 plants tested in vitro cytotoxic effects of plant extracts or compounds (Tables S9 and S10). The main in vitro cytotoxicity testing methods were the brine shrimp toxicity assay and MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide) assay (n = 9 for each). A total 20% of plants showed no significant toxic effects in vitro at the concentrations tested (Table S9) (Table S7).
Another 19 studies tested the toxicity of 20 plants in animal models (Table S11). Mice (65%) and rats (35%) were the models used most frequently. Plant extracts were administered orally (85%) or intraperitoneally (15%). Acacia nilotica L. was the only plant tested both in vitro and in vivo for its toxicity. The majority (60%) of plants tested in vivo had a median lethal dose (LD 50 ) > 2000 mg/kg, while 20% had LD 50 > 5000 mg/kg, indicating low toxicity of the plants tested (Table S9).

| Risk of bias evaluation
All experimental studies, both in vitro and in vivo, were rated as having a probably high risk of bias ( Figure S1 and S2) for not reporting whether blinding of the outcome assessor was performed and whether there was attrition or exclusion of data from analysis. Additionally, six in vitro studies inadequately described their methods; therefore, all domains were categorised as having a probably high risk of bias.
Although all three RCTs showed a low risk of attrition or reporting bias using the RoB 2 tool, none provided information on randomisation, blinding parameters or blinding assessment in their studies ( Figure S3).

| Meta-analyses
Five combinations of plants and parasites met our criteria for metaanalysis (at least three studies evaluating the same plant, the same parasite and the same outcome measure). However, four of these combinations included multiple studies which did not report sufficient variance data to allow for meta-analysis. Thus, we could only perform meta-analyses on one plant/parasite combination.
Hence, we performed two meta-analyses by calculating a pooled IC 50 for the plant and calculating the pooled mean difference in IC 50 compared with the control. Studies by Calzada et al. (2003) andOsuna et al. (2006) had more than one test parameter (different plant parts or compounds). In Osuna et al. (2006), the IC 50 values of six tests were reported only as inactive, and no values were given. Hence, these missing values were imputed using two approaches, as described in the methods, and sensitivity analyses were performed to evaluate the effect of the two imputation approaches. Plants with more than 400 μg/mL IC 50 values are considered inactive against E. histolytica (Nezaratizade et al., 2021). Therefore, this value was used for the second imputation approach.
Results from the sensitivity analyses comparing the overall pooled IC 50 and the overall mean difference using (i) the highest IC 50 value of Osuna et al. (2006) and (ii) the previously established IC 50 value for inactive plants for the parasite are presented in Table 1. The sensitivity analyses showed that imputed outcomes did not significantly affect the pooled effect sizes, and the heterogeneity remained the same.

| Diversity of plants and plant compounds
A total of 507 plant species, most of which belonged to the families Fabaceae, Asteraceae, Combretaceae and Lamiaceae, were pharmacologically evaluated for their antiparasitic potentials against different GI parasites. Fabaceae (beans) is widely distributed and is the thirdlargest land plant family (Christenhusz & Byng, 2016). Asteraceae (daisies) is the largest terrestrial plant family, including over 1600 genera and 25,000 species worldwide (Tähtiharju et al., 2012). Plants of the Combretaceae (white mangrove) family have a wide tropical distribution around the globe, with centres of diversity in Africa and Asia, where they are often used for medicinal purposes (de Morais Lima et al., 2012). Lamiaceae (mints) is an important medicinal family whose members contain a wide range of secondary metabolites, including F I G U R E 4 Forest plot of the mean difference in IC 50 of Lepidium virginicum L. against Entamoeba histolytica compared with the control drug metronidazole, using the maximum IC 50 in Osuna et al. (2006) for missing effect sizes and a random effects model (CI = confidence interval, RE = random effects).
T A B L E 1 Results from meta-analyses using different imputation methods.

Meta-analysis
Imputed outcome F I G U R E 3 Forest plot of effects of Lepidium virginicum L. against the HM1-IMSS strain of Entamoeba histolytica using maximum IC 50 in Osuna et al. (2006) for missing effect sizes and a random effect model (CI = confidence interval, RE = random effects).
phenolic compounds, strong aromatic essential oil and saponins

| Type of studies
Most studies were conducted in vitro rather than in vivo. In vitro experiments provide insights into the cellular response to a plant or drug in a controlled, isolated environment. In vitro screening of plants is cost-effective, less time-consuming than in vivo studies, has a quick turnover of results and allows large-scale screening and does not require the use of animals (Armson et al., 1999). Promising results through a series of in vitro experiments provide a platform for further in vivo studies, which can further evaluate a plant or drug's safety, efficacy and delivery using animal models. In vivo studies are also desirable to assess the pharmacokinetics and pharmacodynamics of the potential compounds and the host immune response to the compounds (Lu & Di, 2020). Compared with in vitro studies, in vivo studies are more realistic and precise. Nevertheless, in vivo testing is costly, more time-consuming and difficult to reproduce due to interanimal differences and pharmacodynamics in the host (Balls, 1994).
Moreover, compounds that are effective in vitro may or may not produce the same in vivo activity (Ferreira et al., 2016;Oliveira et al., 2009). Both in vivo and in vitro methods have vital roles in understanding the efficacy of plants or their derivatives.

| Parasites targeted in studies
E. histolytica and G. duodenalis were the most studied protozoan parasites in vitro and in vivo. E. histolytica can be easily cultured in a laboratory setting in its trophozoite stage. Furthermore, the organism can be genetically manipulated to examine the disease-causing state (Clark & Diamond, 2002). However, the induction of true encystation in the laboratory is difficult (Aguilar-Díaz et al., 2010). Giardia duodenalis is not an intracellular pathogen and does not require host cells for in vitro cultivation. Trophozoites of this species adhere to surfaces and actively replicate via asexual binary fission under optimal conditions and can, therefore, easily be cultivated and maintained in a controlled laboratory as a monoculture (Fink et al., 2020). Additionally, G.
duodenalis is one of the few protozoans that can be induced to complete its lifecycle using laboratory methods (Fink et al., 2020). In contrast to E. histolytica and G. duodenalis, studies focused on the other major GI protozoan parasite of humans, C. parvum, are limited, possibly because of difficulties in culturing this species in vitro (Andrews et al., 2014).
The use of non-parasitic models of GI helminths, such as C. elegans and P. posthuma, was prominent in vitro. This is likely because of the close anatomical and physiological similarities of these nematodes to parasitic nematodes of medical and veterinary importance, their abundance and easy maintenance in the laboratory (Das et al., 2011;Maizels et al., 1993). Not being parasitic species, they lack the process essential for parasitic life cycles, and this may limit the extent to which findings on these models can be generalised to parasitic species.

| Variation in extraction methods
The discovery of possible plant antiparasitic activity typically begins with extractions containing the active molecule(s). Different solvents are used to separate medicinally active molecule(s) from various plant materials (Handa, 2008). Generally, the solvents are chosen based on the polarity of the solute of interest; a solvent with similar polarity to the solute will properly dissolve the solute (Dai & Mumper, 2010 (Boira & Blanquer, 1998). The endogenous factors include the physiological and developmental stages of the plant, as well as the genetic characteristics that regulate their secondary metabolism (Verma & Shukla, 2015). These factors cause differences in the composition of phytochemicals acquired from different plant species as well as in the same plant species.

| Effectiveness of plant extracts
Results of the meta-analyses suggested that L. virginicum is effective for intestinal protozoan parasite E. histolytica. The entire plant, root and isolated compounds benzyl glucosinolate and β-Sitosterol of L. virginicum inhibited the in vitro growth of E. histolytica trophozoites with promising IC 50 values (Calzada et al., 1998(Calzada et al., , 2003Osuna et al., 2006). This aligns with its ethnomedicinal use in Ayurveda and Traditional Mexican Medicine for treating diarrhoea, dysentery and worm infections (Calzada et al., 2003;Chevallier, 1996;Khare, 2004).
Lepidium virginicum belongs to the Brassicaceae (mustard) family. It is a native plant in North America and can be found elsewhere as an introduced species (Calzada et al., 2003). All parts of the plant are edible, eaten raw or cooked. This plant has promise as an antiparasitic agent and would be a good candidate for further research to explore its in vivo activity and toxicity profiles.

| Toxicity of plants and plant compounds
The belief that medicinal plants are safer than synthetic drugs has caused remarkable growth in human exposure to plant products (Haq, 2004). This has also led to a resurgence of scientific interest in the biological effects of plants (Cragg & Newman, 2013;Kinghorn et al., 2011). Plants produce phytochemicals to protect themselves from various environmental threats, such as herbivores and microorganisms (Wink, 1988). Some phytochemicals are natural toxins, and these toxic substances, when ingested, can be potentially harmful to human health (Omaye, 2004 We could perform only two meta-analyses due to the lack of homogeneity across individual studies and missing variance data, even if there were homogenous studies. In addition, inadequate reporting of data (particularly variance data and IC 50 values of agents deemed to be inactive) created challenges for our meta-analyses.

| CONCLUSIONS AND RECOMMENDATIONS
This systematic review demonstrates that a rich diversity of plant species have been pharmacologically evaluated for their antiparasitic potential against human GI parasites. The findings highlight the importance of pharmacokinetics and pharmacodynamics studies for the industrial development of new antiparasitic compounds. We provide comprehensive resources listing the plant and plant compounds that have shown promise as antiparasitic agents (Tables S1, S2, S5, S6 and S7), those that have not (Tables S3, S4 and S8) and those that have been tested for toxicity (Tables S9-S11). These tables may help researchers identify appropriate plant and plant compounds to investigate further for their antiparasitic potential and toxicity.
Based on our findings and, in particular, weaknesses identified in the literature, we present various recommendations for future studies investigating the antiparasitic effects of plant and plant compounds against human GI parasites.
We believe that future research should • provide chemical or botanical characterisation data of the plant to enhance the reproducibility and accuracy of the findings; • evaluate the toxicity measures of plant extracts or compounds before evaluating their antiparasitic effects to ensure the antiparasitic potential is not due to their toxic effects; • focus on in vivo evaluation of successful in vitro candidates since many promising in vitro candidates do not appear to have been assessed in vivo; • implement animal research reporting guidelines for reporting methods to maximise the quality and reliability of research; and • report the methodology of the studies in detail and provide complete statistics of the outcome (e.g. SEM, SD, SE or CI) for precision and to enable future researchers to use the data. writingreview and editing. Amber Beynon: Data curation; methodology; writingreview and editing. Amanda Ash: Supervision;