For the development of this study, all original indexed articles, written in English, Portuguese, and Spanish, which reported data about the activities of extracts, fractions, and isolated compounds from the genus Zanthoxylum against Leishmania, Plasmodium, Trypanosoma cruzi, and Trypanosoma brucei on preclinical experiments (in vitro and/or in vivo) were included. Review articles, book chapters, case studies, and articles about genus activity against protozoan vectors and activities of synthesized compounds were excluded (Figure 1). Initially, a protocol was developed to carry out the systematic review, and this was approved by Prospero (CRD42020200438).

The electronic databases PubMed, Virtual Health Library (VHL), Scopus, Science Direct, and Web of Science, all free access databases, were used to search for articles. The search descriptors were combinations of terms found in the titles of medical subjects (Mesh) and Descriptors in Health Sciences (DeCs). The date of the last search was May 2020.

In PubMed (pubmed.ncbi.nlm.nih.gov), there was mainly North American literature, and the descriptors used were as follows: (zanthoxylum) OR (zanthoxylums)) OR (xanthoxylum)) OR (xanthoxylums)) OR (fagaras, Zanthoxylum)) AND ((((Leishmaniasis) OR (Chagas disease)) OR (malaria)) OR (African trypanosomiasis)). In the BVS (bvsalud.org), there was a predominance of literature from Latin America when the following terms were used: (((leishmaniose) OR (doença de chagas) OR (malaria) OR (tripanossomíase Africana))) AND (tw: ((zanthoxylum)) OR (xanthoxylum).

In the Scopus database (scopus.com), which includes several knowledge areas, we used the following descriptors: (zanthoxylum OR xanthoxylum OR “fagaras, Zanthoxylum”) AND ALL (leishmaniasis OR “Leishmania Infection” OR “Chagas disease”" OR “Trypanosoma cruzi Infection” OR malaria OR “Plasmodium Infection” OR “African trypanosomiasis” OR “African Sleeping Sickness").

Operated by the Anglo-Dutch publisher Elsevier, Science Direct (sciencedirect.com) was also included in the research and the descriptors were as follows: (zanthoxylum OR xanthoxylum) AND (leishmaniasis OR Leishmania Infection OR Chagas disease OR Trypanosoma cruzi Infection OR malaria OR Plasmodium infection OR African trypanosomiasis). Finally, we included Web of Science (webofscience.com) as a multidisciplinary base and the research adopted the descriptors (zanthoxylum OR xanthoxylum OR fagaras, Zanthoxylum”) AND (leishmaniasis OR chagas disease” OR malaria OR african trypanosomiasis").

The articles were selected by two independent examiners, based on the reading of the title and abstract, with a third final examiner. Potentially eligible articles were read in full.

Because the selected studies may present risks of bias, two instruments were used: the guidelines for pre-clinical in vitro reports on dental materials, developed by Faggion et al. (2012) modified to meet the criteria of in vitro studies (Table 1), and SYRCLE, developed by Hooijmans et al. (2014) to assess the risk of bias of in vivo studies (Table 2). The studies were analyzed in pairs by two examiners, with a third evaluator when there was disagreement.

Based on the material developed by Faggion et al. (2012), the risk of bias was classified into numbers, with the value one representing a low risk of bias; 0.5, potential risk of bias; and 0, risk of bias, as shown in Table 1. This instrument was chosen because it is an easy-to-handle tool with a high intra-examiner agreement. The analysis was based on the following aspects: 1) abstract; 2) introduction, background, and objectives; 3) methods and intervention; 4) results; 5) sample size; 6) randomization: sequence generation; 7) allocation concealment; 8) implementation; 9) blinding (10) statistical methods; 11) results and estimates 12) discussion; 13) other information (financing); and 14) protocol.

The tool developed by Hooijmans et al. (2014) was created specifically for intervention studies in animals and is based on the Cochrane tool. The risk of bias was classified into specific characters, where N (not relevant to the item); S (showed relevance to the item); and M (presented medium relevance to the item). The analysis was based on the aspects (1; 2; 3) selection bias; (4; 5) performance bias (6; 7) detection bias; 8) friction bias; 9) random bias; and (10) other sources of bias (Table 2).

Regarding the activity classification, the parameters of Dolabela et al. (2015) for the genus Plasmodium and those of Mota et al. (2015) for the genus Leishmania were used, and the following parameters were adopted to evaluate the activity against Trypanosoma cruzi and Trypanosoma brucei: IC₅₀ < 10 μg/ml, good activity; IC₅₀ of 10–50 μg/ml, moderate activity; IC₅₀ 50–100 μg/ml, low activity; and IC₅₀ > 100 μg/ml, inactive. For the selectivity index (IS), the one calculated from the ratio between cytotoxicity for macrophages (CC₅₀) and activity against amastigotes (IC₅₀) was used. IS values >20.0 indicate that the sample tested was more toxic to the parasite than to the host cell. IS values <20.0 demonstrate toxicity of the compound, adapted as described by Don and Ioset (2014).

Variables of interest for in vitro studies (first author, year of study, genus, species, part of the plant used, 50% inhibitory concentration—IC₅₀, evolutionary form, protozoan species, 50% cytotoxic concentration—CC₅₀, cell lineage, selectivity index, and activity classification) were transferred to tables in Word (Tables 3, 5, 7, 9).

Variables of interest for in vivo studies (first author, year of study, genus, species, part of the plant used, % parasitemia, % chemosuppression, % inhibition of the protozoan, protozoan species, animal species, 50% cytotoxic concentration (CC₅₀), cell lineage, and selectivity index) were transferred to tables in Word (Tables 4, 6, 8).

In the search, 1,491 articles were identified, of which only 166 met the inclusion criteria. In the screening phase, the authors evaluated the titles of these 166 articles. Of these, 101 articles were excluded due to duplicity, and therefore, 65 articles were chosen for reading. After analyzing the abstracts, eight articles were excluded because they did not study the genus Zanthoxylum, four because they did not generate the IC₅₀, four because they did not present numerical values, one because the tested parasite did not infect humans, one because it did not portray protozoa, one because it was not in a document available for reading without open access, one for being ethnobotanical article, and one review article, totaling 21 articles excluded. Thus, 44 original articles were included in this systematic review (Figure 1).

In the introduction and initial part of the methodology, all articles met the verification completely or partially, noting that the first items are in accordance with the methodology adapted from Faggion Jr et al. (2012). However, the verification in the final part of the methodology is totally different from the initial one, as none of the articles meet the guideline verification criteria, and this is a warning as to how the methodologies of these in vitro articles are being managed (Table 1).

However, based on the results and discussion, it is possible to notice again the improvement in the verification of these articles, and they fully or partially meet this verification. However, in the last item (protocol), none of the articles meet the verification criteria. This means that the articles did not expose their protocols, which hinders the reproducibility of this work (Table 1).

In the end, to have a quality overview of these articles, a score was generated according to the items the article presented during the methodology verification, and after that, we verified that the studies by Aratikatla et al. (2017), Bertani et al. (2005), Bouquet et al. (2012), Cebrián-Torrejón et al. (2011), Enciso et al. (2014), Gansane et al. (2010), Gessler et al. (1994), Kirira et al. (2006), Kamanzi Atindehou et al. (2004), Ohashi et al. (2018), Penali et al. (2007), Rukunga et al. (2009), Sandjo et al. (2016), and Tchinda et al. (2009) and obtained the highest score (score 9 out of 14) among the 37 articles evaluated by this methodology (Table 1).

None of the articles evaluated by Hooijmans et al. (2014) met the satisfaction criteria in selection, performance, and detection biases. Only the studies by Bertani et al. (2005), Musila et al. (2013), and Muganga et al. (2014) showed satisfaction in the assessment used in the reporting bias (Table 2).

An extensive literature review was carried out to identify whether extracts, fractions, and pure substances from the genus Zanthoxylum presented in vitro and in vivo activity against protozoa of the genus Plasmodium, Leishmania, and Trypanosoma. The results can be seen in Tables 3, 4, 5, 6, 7, 8, 9.

One study evaluated the antiplasmodic activity against 12 alkaloid fractions from the bark of the Z. zanthoxyloides stem against resistant strains of P. falciparum. Fractions 1–5, 10, and 12 were active and non-cytotoxic to chronic myeloid leukemia strain (Table 3; Gansane et al., 2010). The ethyl acetate extract from the stem bark of Zanthoxylum chalybeum was active against sensitive and resistant strains of P. falciparum (Table 3; Adia et al., 2016).

Ethanolic ethyl acetate and aqueous extracts from the Z. chalybeum root bark showed activity against the same resistant strains (Table 3; Gessler et al., 1994). In addition, the aqueous extract from the Z. chalybeum root bark was active against resistant strains of the parasite, and the methanol extract was active against resistant and sensitive strains (Table 3; Muthaura et al., 2015).

From Z. rhoifolium, one alkaloid was isolated, and the alkaloid nitidine was considered active against the resistant strain of the P. falciparum (Table 3; Jullian et al., 2006). The alkaloid nitidine isolated from Z. rhoifolium was active against two resistant strains and a sensitive strain of P. falciparum, and it was not cytotoxic against monkey kidney fibroblast cells (Table 3; Bouquet et al., 2012).

The hexane extract from the bark of Z. heitzii was active on sensitive strains of P. falciparum. The alkaloids dihydronitidine, pellitorine, heitziquinone, caryophyllene oxide, rhoifoline B, and isoarnottianamide were isolated from the same species. From them, dihydronitidine and pellitorine showed activity against P. falciparum strains.

Finally, for in vitro studies, the syncarpamide, isolated from Z. syncarpum, was active in resistant and sensitive strains of P. falciparum, being non-cytotoxic against the normal monkey kidney fibroblast cells with high SI (Table 3; Aratikatla et al., 2017).

Regarding in vivo studies, percentage parasitemia and percentage chemosuppression were determined using the formula % parasitemia = (total number of parasite cells/total number of cells) x 100% and chemosuppression = [(negative control parasitemia) − (parasitemia with drug)]/negative control parasitemia (Hilou et al., 2006). After oral treatment with 200 mg/kg of methanol extract from Fagara zanthoxyloides in albino Wistar mice infected with Plasmodium berghei, the level of parasitemia was 33%, chemosuppression was 82.37%, and the inhibition rate of the parasite load was 68.39%. After treatment with a dose of 400 mg/kg of the same extract, the level of parasitemia was 41.50%, chemosuppression was 90.75%, and the inhibition rate of the parasite load was 84.84%. After treatment with the highest dose (600 mg/kg) of the extract, the parasitemia was 44.15%, with a chemosuppression of 95.95%, and inhibition of the parasite load of 92.67%. Furthermore, the non-lethality of the extract was confirmed, even at the highest dosage (5,000 mg/kg), having a CC₅₀ of 28.21 ± 1.30 μg/ml (Table 4; Enechi et al., 2019).

An oral treatment with 100 mg/kg of aqueous and methanol + chloroform extract obtained from the bark of the Zanthoxylum chalybeum stem was administered to Swiss albino mice infected with Plasmodium berghei; parasitemia of 17.56% and chemosuppression of 44.93% were found after treatment, and parasitemia of 23.09% and chemosuppression of 27.56% after treatment with methanol + chloroform extract were also found. The two extracts were not cytotoxic (CC₅₀: 268.28 and 25.78 μg/ml, respectively) at a dosage >1,000 μg/ml (Table 4), and for each specific blood smear for a given mouse, four magnification fields were observed, and the number of parasitized cells and the total number of cells in the magnification field were recorded. The data obtained were used to determine the percentage parasitemia and percentage chemosuppression in each mouse (Musila et al., 2013).

Finally, the inhibition of the parasite load in Swiss mice infected with P. yoelii after oral treatment with 715 mg/kg of the aqueous extract from Z. rhoifolium was 78.29% (Table 4; Bertani et al., 2005). In this study, there was no calculation of parasitemia.

In in vitro studies, the ethanolic extract and the hexane fraction from the bark of the Z. rhoifolium stem showed activity against Leishmania amazonensis promastigotes in 24, 48, and 72 h of treatment (Table 5; Melo-Neto et al., 2016). In the same study, the alkaloids chelerythrin and nitidine, isolated from Z. rhoifolium, showed activity against L. amazonensis amastigotes (Table 5; Castillo et al., 2014).

The dichloromethane extract from the bark of the Z. zanthoxyloides stem showed activity against L. donovani promastigotes; it was not cytotoxic in Jurkat cells (Table 5; Dofuor et al., 2019a).

The ethanolic extract and the alkaloid fraction from the bark of Z. monophyllum showed activity against L. major and L. panamensis promastigotes (Table 5; Enciso et al., 2014). Moreover, the ethanolic extract and the n-hexane fraction from Z. armatum fruits showed activity when tested against Leishmania major promastigotes (Table 5; Alam and Najam us Saqib, 2017).

The hydroethanolic and the aqueous extracts from the bark of Z. juniperinum showed activity against the promastigote form of Leishmania spp. (Table 5; CHINCHILLA-(Chinchilla-Carmona et al., 2012). In addition, the alkaloid buesgenine isolated from Z. buesgenii showed activity against amastigotes of L. amazonensis (Table 5; Sandjo et al., 2016). Also, the syringoresinol lignan, isolated from Z. tingoassuiba, was active against L. amazonensis promastigotes (Table 5; Costa et al., 2018).

Regarding in vivo studies, only isolated compounds of the genus were tested against Leishmania. The antileishmanial activity of canthin-6-one and 5-methoxycanthin-6-one isolated from Zanthoxylum chiloperone was evaluated in BALB/c mice infected with Leishmania amazonensis strains at a dose of 10 mg/kg by different routes (oral and intralesional). After oral treatment with canthin-6-one and 5-methoxycanthin-6-one, the parasite load inhibition rate was 171.4% and 68.4%, respectively, and chemosuppression of 139.6% and 27.7%, respectively. After treatment via the intralesional route, the parasitic inhibition rate was 77.6% and 21.6%, respectively, and the level of chemosuppression was 15% and 124.2%. It did not present toxicity, and its lethal dose was >400 mg/kg intraperitoneally (Table 6; Ferreira et al., 2002).

The alkaloids fagaridin and chelerythrine isolated from Zanthoxylum rhoifolium were also evaluated to assess their antileishmanial activity in BALB/c mice infected with L. amazonensis strains treated with a dose of 5 mg/kg by intralesional injection. After treatment, the parasitemia was 3.96×10⁶ ± 0.48 for fagaridin and 6.75×10⁶ ± 0.29 for chelerythrine; the inhibition rate of the parasite load was 59.6% and 29%, respectively (Table 6; Castillo et al., 2014). In none of the in vivo studies was there an analysis of parasitemia in percentage (%), only the mean was presented.

The hexane extract from the bark of Zanthoxylum naranjillo showed high activity against different strains of Trypanosoma cruzi trypomastigotes (Table 7; Bastos et al., 1999), followed by the hexane fraction of Z. minutiflorum with activity against T. cruzi trypomastigotes (Table 7; Mafezoli et al., 2000). The lignan compound syringoresinol isolated from the species Z. tingoassuiba was the isolated compound that showed the strongest activity against T. cruzi epimastigotes (Table 7; Costa et al., 2018). No studies examined compound activity against cells in the amastigote phase.

In in vivo studies, the crude extract of the stem of Zanthoxylum chiloperone showed the strongest antitrypanosome activity, being evaluated at a dose of 50 mg/kg orally and subcutaneously in BALB/c mice infected with Trypanosoma cruzi. After oral treatment, a parasitemia of 0.7×10⁴ ± 0.8 was found, and a parasitemia of 2.7×10⁴ ± 1.7 was found after subcutaneous administration (Table 8; Ferreira et al., 2007).

In addition, the alkaloid canthin-6-one, also isolated from the species Zanthoxylum chiloperone, was tested in BALB/c mice infected with T. cruzi strains at a dose of 5 mg/kg, orally and subcutaneously. After exposure to oral treatment, the parasitemia was 0.1×10⁴ ± 0.2, and after subcutaneous treatment, the parasitemia was 3.3×10⁴ ± 2.7, representing the compound with the strongest in vivo activity (Table 8; Ferreira et al., 2007).

The dichloromethane extract obtained from the root of Z. zanthoxyloides showed the strongest activity against T. brucei (Table 9; Dofuor et al., 2019a). Moreover, the alkylamide tortozanthoxylamide, isolated from Z. zanthoxyloides, was the only isolated compound of the genus that was found in tests against T. brucei showing activity against the protozoan, moderate cytotoxicity in rat macrophages, and high SI (Table 9; Dofuor et al., 2019b).