Medicinal Plants from North and Central America and the Caribbean Considered Toxic for Humans: The Other Side of the Coin

The consumption of medicinal plants has notably increased over the past two decades. People consider herbal products as safe because of their natural origin, without taking into consideration whether these plants contain a toxic principle. This represents a serious health problem. A bibliographic search was carried out using published scientific material on native plants from Mexico, Central America, and the Caribbean, which describe the ethnobotanical and toxicological information of medicinal plants empirically considered to be toxic. A total of 216 medicinal plants belonging to 77 families have been reported as toxic. Of these plants, 76 had been studied, and 140 plants lacked studies regarding their toxicological effects. The toxicity of 16 plants species has been reported in clinical cases, particularly in children. From these plants, deaths have been reported with the consumption of Chenopodium ambrosioides, Argemone mexicana, and Thevetia peruviana. In most of the cases, the principle of the plant responsible for the toxicity is unknown. There is limited information about the toxicity of medicinal plants used in Mexico, Central America, and the Caribbean. More toxicological studies are necessary to contribute information about the safe use of the medicinal plants cited in this review.


Introduction
The use of herbal medicine has increased around the world due to its presumptive efficiency, availability, and general acceptance. Approximately 80% of the general population, especially in developing countries, uses medicinal herbs for primary health care [1,2]. Worldwide, the interest in medicinal plants by patients has increased over the past two decades. The global market for medicinal plants and plantderived drugs in 2015 was estimated at 25.6 billion dollars and is expected to rise to 35.4 billion dollars in 2020 [3]. This clearly indicates that the consumption of medicinal plants is a current topic of interest. Despite the high consumption of medicinal plants and related products, their toxicity remains to be evaluated. In addition, many medicinal plants require scientific evidence for their medicinal use, particularly those that are sold as food supplements.
Some medicinal plants might trigger undesirable side effects for human health because of (a) pharmacodynamic interaction with prescribed drugs, (b) intrinsic effects, (c) pharmacokinetic interaction with prescribed drugs, and (d) the presence of contaminants and/or pathogenic microorganisms. Other factors that impact the toxicity of medicinal plants in humans include the age of the patient, nutritional status, and the presence of chronic diseases. The concentration of toxic metabolites in plants is determined by the season 2 Evidence-Based Complementary and Alternative Medicine of collection, nutrients in the soil, and growth stage, among others [4].
The main reasons for focusing this review on medicinal flora with supposed toxicological effects from Mexico, Central America, and the Caribbean are as follows: (a) the ancient importance of traditional medicine in this region, (b) their great biodiversity, and (c) the current use of herbal products. This review will be useful for physicians, toxicologists, pharmacologists, and general audiences. We have tried to describe in detail some toxic symptoms reported with the consumption of the medicinal plants covered in this review.

Methodology
A bibliographic search was conducted from July 2016 to May 2017 of published scientific material on native plants from Mexico, Central America, and the Caribbean that describes the ethnomedical and toxicological information for medicinal plants reputed to be toxic for humans. The following keywords were searched in different scientific databases: plant extract, toxicity, Mexico, and Central America. Additional data were acquired from undergraduate and postgraduate theses, as well as published and electronic books. The admittance criteria for the selection of scientific information in this review were as follows: (i) plants native to North and Central America and the Caribbean, (ii) plants used for medicinal purposes with or without toxicological studies, (iii) plants with experimental reports on their in vitro and/or in vivo toxicity, (iv) plants thought to be toxic for humans, (v) studies where the concentrations were presented as weight/volume relationship in international units (mg/ml, g/ml), (vi) studies where the doses were presented as weight/weight relationship in international units (mg/kg, g/kg), and (vii) plants with information obtained from a clear source. Scientific studies reporting the combination of plant extracts were excluded. Medicinal plants considered toxic were classified into two categories: (1) plants with toxicological evidence reported in a scientific source and (2) plants without toxicological evidence. All plant names and their distributions were confirmed at the Missouri botanical garden [73]. Many of the medicinal plants cited in this review have no common name in English. Therefore, the common names were given in Spanish (Table 1).

Dosages.
In most of the cases, the dose for the induction of toxic effects by medicinal plants is not indicated. Usually, consumers of medicinal plants believe that increasing the consumption of these products will increase the efficacy of the treatment. In these cases, the daily dosage is exceeded, which triggers toxicity. For instance, the roots of Ipomoea purga, a purgative agent, are used at a dose of 2 g/L/day. Administration of higher doses induces vomiting and abdominal pain [15]. Fresh leaves of Prunus serotina, used for the treatment of cough, or Zanthoxylum fagara, an anxiolytic agent, each must be consumed in a maximum quantity of five leaves in 250 ml of water per day. Higher doses produce spasms and nausea [26]. Approximately 5 mL of an infusion of Picrasma excelsa (10 g/L) should be administered three times per day. Higher doses induce hypotension. This infusion should not be prepared with ethanol and orally administered. If a person consumes the hydroalcoholic infusion, the consequences could be lethal [35]. The maximum consumption of Manilkara zapota seeds should be 10 seeds per day. A higher consumption of these seeds might induce vomiting and gastroenteritis [26]. On the other hand, Sosa-Gómez [35] recommends the preparation of an infusion using approximately 1-3 g Argemone mexicana leaves in 1 L of water. This infusion should be taken 3 times per day. A higher dose might induce immobilization.
Studies analyzing the range of doses considered safe for human consumption remain to be performed. The use of natural products needs scientific evidence to corroborate the medicinal uses attributed to different plant species. Many medicinal plants sold as "food supplements" lack warnings if the suggested dosage is exceeded.

Toxic Principles.
In some cases, the toxic principle is known. For instance, it is reported that cefalatin, the main toxic compound in Cephalanthus occidentalis bark, induces vomiting, anemia, and seizures, among other toxic effects. Similarly, hederagenin is the main toxic compound in Clematis dioica, which is a caustic substance [4]. Monocrotaline is the compound responsible for the toxic effects in Crotalaria sagittalis. Cianhidric acid, one of the most toxic compounds in plants, is found in Crescentia cujete fruit, Phaseolus lunatus whole plant, and Prunus serotina leaves and seeds [4]. In Phaseolus lunatus, the concentration of cianhidric acid ranges 6.8-533 mg/kg dw [74,75]. There is limited information on the major toxic compounds cited in this review. Therefore, the identification of toxic principles in medicinal plants is necessary.
Evidence-Based Complementary and Alternative Medicine 3  Evidence-Based Complementary and Alternative Medicine 5

In Vitro Studies.
The Artemia salina (brine shrimp) bioassay has been widely used for the analysis of acute toxicity in vitro. Although there are no range values to consider an extract or compound as toxic in the brine shrimp test, vincristine, the positive control for toxicity, has a lethal concentration 50 (LC 50 ) = 0.91 g/ml [76]. Considering this value, plant extracts or compounds with LC 50 values 1000fold higher than vincristine could be considered nontoxic.
The following plant extracts have been tested for their in vitro toxicology using the brine shrimp test and had LC 50 values higher than 1000 g/ml. The ethyl acetate fraction of Solanum nigrescens aerial parts [77], the ethanol extract of Ambrosia peruviana whole plant [78], the aqueous extract of Jatropha gossypifolia aerial parts [79], the methanol extract of Jatropha dioica leaves [80], the aqueous extract of Cnidoscolus urens whole plant [79], the ethanol extract of Crescentia cujete fruits [81], the aqueous extract of Enterolobium cyclocarpum bark [82], and the ethanol extract of Cordia dentata leaves and their fractions [83].
The plant extract and compounds that could be considered dangerous (LC 50 = 100-1000 g/ml) include the following: methanol and hexane proportions derived from a hexane extract of Gymnosperma glutinosum aerial parts [84], the ethyl acetate extracts of Monstera deliciosa branches [85], and icosandrin, a cyclic homoflavonoid isolated from Phytolacca icosandra [86].
The plant extract and compounds that could be considered toxic (LC 50 = 10-100 g/ml) include the following: the ethyl acetate extracts of Monstera deliciosa leaves [85], the ethanol extract of Solanum americanum fruits [87], the ethanol extract of Scoparia dulcis aerial parts [76], the methanol extract of Enterolobium cyclocarpum leaves [88], the ethanol extract of Pimenta dioica leaves [89], and the hydroalcoholic extract of Sanguinaria canadensis whole plant [90]. None of the plant extracts or compounds included in this review were considered highly toxic (LC 50 < 10 g/ml).

Cytotoxicity.
Other plant extracts and their compounds have been tested in other in vitro models, including cytotoxicity test in nontumorigenic cells, genotoxicity using the comet assay on lymphocytes, and the mutagenic test using lymphocytes or Salmonella spp. The positive controls for cytotoxicity in nontumorigenic cells include cisplatin and Taxol. These compounds have inhibitory concentration 50 (IC 50 ) values ranging from 0.1 to 4 g/ml [91]. Some plants extracts have been reported to lack cytotoxic effects (IC 50 > 250 g/ml) in nontumorigenic cells. These include the ethanol extract of Phoradendron serotinum leaves tested on peripheral blood mononuclear cells [91], the aqueous extract of Cnidoscolus chayamansa leaves on baby hamster kidney (BHK) cells [92], and the aqueous extract of Enterolobium cyclocarpum bark assayed on 3T3 murine preadipocytes [82]. Additionally, the ethanol extract of Equisetum hyemale aerial parts evaluated on rabbit corneal fibroblasts (SIRC) [93], the methanol extract of Enterolobium cyclocarpum leaves evaluated on Vero cells (obtained from kidney epithelial cells extracted from the African green monkey (Cercopithecus aethiops) [88], and the diterpene ent-kaur-16-en-19-oic acid, obtained from Annona cherimola, tested on rat embryo primary striatal cultures [94]. On the other hand, the hydroalcoholic extract of Hura crepitans leaves had an IC 50 = 107.7 g/ml in lung fibroblasts [95].
For genotoxicity, the ethyl acetate/n-hexane extract of Zinnia peruviana aerial parts tested using 5 and 20 mg/ml extracts showed genotoxic effects in PBMC compared to the positive control of copper sulfate (1%) [100]. A butanol fraction of Urera baccifera roots at a 1.8 mg/g concentration of oxalic acid decreased leukocyte number significantly and increased cell death and DNA damage in primary cultures of leukocytes in comparison to the control treatment [101]. The methanolic extract of Psittacanthus calyculatus aerial parts (200 and 400 mg/kg i.p.) did not induce chromosomal damage in peripheral blood erythrocytes obtained from mice after 72 h of exposure [102]. An ethanol extract of Heliopsis longipes roots (3-100 mg/kg p.o.) did not produce genotoxic or cytotoxic effects on peripheral blood mononuclear cells obtained from mice 24-96 h after administration [103]. Parthenin, isolated from Parthenium hysterophorus, showed genotoxic effects at 4-31 mg/kg i.p. in micronuclei in mouse peripheral blood after 48 and 72 h of treatment [96]. 50 ). The guideline 423 of the Organization for Economic Cooperation and Development (OECD) establishes that substances with an LD 50 < 5 mg/kg are highly toxic, whereas LD 50 values from 5 to 50 mg/kg are very toxic, LD 50 values from 50 to 300 mg/kg are toxic, LD 50 values from 300 to 2000 mg/kg are dangerous, and LD 50 values higher than 2000 mg/kg are not dangerous [104].

Biochemical and Hematological Parameters.
Treatment with plant extracts in short-term studies have effects on biochemical and hematological parameters, as well as the levels of the hepatic enzymes alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP). An aqueous extract of Larrea divaricata aerial parts (0.5-200 mg/kg i.p.) did not affect the levels of ALT in mice after 2 days [153]. Aqueous extract of Karwinskia humboldtiana fruits (1250 mg/kg p.o.) administered to rats for 3 days increased the levels of hepatic enzymes compared to the untreated group [154]. An aqueous extract of Passiflora edulis (30 mg/kg p.o.) did not affect motor coordination in mice or change the biochemical measurements in serum after 4 days [155]. -Solamargine (15-35 mg/kg i.p.) did not affect hematological parameters or the levels of hepatic enzymes in rats after 5 days [148], whereas an aqueous extract of Karwinskia humboldtiana fruit (5000 mg/kg p.o.) in rats for 5 days induced weight loss (15%) in rats, as well as toxicity in the pancreas [156]. An aqueous extract of Passiflora edulis leaves (100-400 mg/kg p.o.) did not affect organ body weight or hematological parameters but decreased the levels of ALT in rats after 7 days of treatment at all doses [126]. A methanol extract of R. tetraphylla leaves (1000 mg/kg p.o.) decreased body weight change and food consumption and increased total bilirubin in rats after 7 days [157]. An aqueous extract of K. humboldtiana (1000-2000 mg/kg p.o.) administered for 7 days in rats induced alterations in membrane fluidity and ATPase activity in liver submitochondrial particles [158]. An ethanol extract of Euphorbia hirta leaves (60.4-483 mg/kg i.p.) [159] and a chloroform fraction from an ethanol extract of Tagetes erecta flowers (200-400 mg/kg p.o.) [133] did not affect hematological or biochemical parameters in rats after 14 days. An aqueous extract of Euphorbia hirta whole plant administered at a single dose of 2000 mg/kg p.o. significantly decreased the levels of ALP and ALT after 14 days in broiler chickens [160]. Lobeline (5-10 mg/kg i.p.) did not affect the levels of AST, ALT, ALP, and LDH after 4 days of exposure [99].

Toxicity to Reproduction and Pregnancy.
-Solamargine (15-35 mg/kg i.p.) did not affect the number of spermatozoa or the weight of the testicles and epididymis of male rats after 24 h of treatment [148]. Jervine (70-300 mg/kg p.o.), a steroidal alkaloid found in Veratrum californicum, administered on days 8-10 of gestation induced malformations in the offspring, including isolated cleft palate, mandibular micrognathia, and limb malformations in C57BL/6J and A/J mice [161]. The administration of an aqueous extract of Ruta chalepensis leaves (10 mg/kg p.o.) to pregnant rats from day 9 to day 17 of gestation decreased the uterine weight, the number of live fetuses, and the fetal weight.
The number of fetal resorptions was also increased, and the fetuses showed skeletal malformations [162]. Additionally, an aqueous extract of R. chalepensis leaves (0.8 and 1.6 g/kg p.o.) administered to mice from day 1 to day 14 post coitum caused perinatal changes in mice such as righting reflex, cliff avoidance, and swimming ability, among others [163].

Dermal
Tests. An aqueous extract of Pouteria sapota seeds lacked dermal irritation in rabbits and showed mild reversible eye irritation in rabbits [122]. Jatropha multifida sap did not induce skin lesions in rats after 14 days of treatment [164]. A diethyl ether extract of Jatropha multifida showed the presence of 16-hydroxyphorbol. This compound showed an irritant dose 50 (ID 50 ) of 0.05 g/ear [165].

Biochemical and Hematological Parameters.
Some plant extracts and their active compounds have been tested for their effects on biochemical and hematological parameters in rodents for at least 18 days of exposure. A histological study has also been included in some reports. A hydroalcoholic extract of Senna occidentalis aerial parts (100-2500 mg/kg p.o.) [135] and a hydroalcoholic extract of Sapindus saponaria leaves (44.76 mg/kg p.o.) and fruits (45.0 mg/kg p.o.) [166] did not change the biochemical profile or hematological parameters or alter body weight and organ weight for 30 days in rats. An ethanol extract of the pod of Plumeria rubra (50-200 mg/kg p.o) [167] and an aqueous-methanol extract of Ceiba pentandra leaves (250 and 500 mg/kg p.o.) [117] did not alter hematological or biochemical parameters in rats and mice, respectively, after 21 days [167]. A chloroform-methanol extract of Cnidoscolus chayamansa leaves (1000 mg/kg p.o.) [131] and an ethanol extract of Moussonia deppeana aerial parts (1000 mg/kg p.o.) [107] did not affect biochemical or hematological parameters in mice after 28 days of daily administration. In addition, histological examinations of the spleen, kidney, and liver showed no abnormalities. Capsaicin (5-100 mg/kg p.o.) [168], obtained from Capsicum annum, and lutein and lutein ester (4-400 mg/kg p.o.) [136], obtained from Tagetes erecta, did not affect hematological or biochemical parameters, growth, food consumption, or body weight in mice and rats, respectively, after 28 days. An aqueous extract of Rivina humilis fruits (2500 and 5000 mg/kg p.o.) administered daily for 35 days showed no changes in the hematological profile or in the relative organ weight, whereas the same extract administered daily for 90 days at 0.5-2 g/100 g in a powdered diet did not affect hematological parameters, biochemical determinations, or the levels of hepatic enzymes [132].
In contrast, some plant extracts have altered biochemical and/or hematological parameters. An aqueous extract of Abrus precatorius leaves (400-1600 mg/kg p.o.) was administered to rats for 18 days. Only the highest dose (1600 mg/kg p.o.) decreased levels of hematological parameters and increased the levels of hepatic enzymes [169]. An aqueous extract of Scoparia dulcis leaves (250-500 mg/kg p.o.) showed mild vascular and portal congestion in the heart and the liver, respectively, of rats treated daily with this extract for 30 days. Nevertheless, there were effects in the lungs and testis [120]. A methanol extract of Rauvolfia tetraphylla leaves did not affect hematological parameters. However, a significant decrease in the total bilirubin and glucose levels was observed in the mice treated at 100 and 300 mg/kg, with a significant increase in triglycerides at doses of 10-300 mg/kg after 28 days in mice [130]. An ethanol extract of the aerial parts from Jatropha gossypifolia (135 mg/kg p.o. or higher doses) reduced the activity in the central nervous system and showed hepatotoxicity, pulmonary damage, and digestive disturbances in rats over 13 weeks of treatment [170]. The lethality was 46.6% and 13.3% among male and female rats under the highest tested dose (405 mg/kg), respectively [170].

Toxicity during Reproduction and Pregnancy.
Yao et al. [171] reported that an aqueous extract of Hydrastis canadensis (1.86 g/kg p.o.) containing 9.6 mg/ml of berberine and 8.4 mg/ml of hydrastine did not affect fetal development in pregnant rats over 20 days of treatment. A Prosopis juliflora seedcase added at 70% to the diet of pregnant rats resulted to be teratogenic (13-fold) compared to the untreated group [172]. Aqueous and ethanol extracts of Plumeria rubra pods (200 mg/kg p.o.) had 51% and 100% abortifacient activity, respectively, in female albino rats from day 11 to day 15 of pregnancy [173]. The hydroalcoholic extract of Lantana camara leaves (1000-7000 mg/kg p.o.) administered during premating, mating, pregnancy, and lactation (56 days in total) in rats did not affect mating, pregnancy, delivery, and live birth. Nevertheless, the two highest doses tested (3000 and 7000 mg/kg p.o.) produced an increase in the resorption rate and parallel increases in the postimplantation loss index, as well as embryotoxicity characterized by skeletal abnormalities [174].

Carcinogenicity.
Only one plant extract has been tested for its carcinogenic effects. Rats (treated with doses ranging 136-1175 mg/kg p.o.) and mice (treated with doses ranging 375-3275 mg/kg) received an aqueous extract of Hydrastis canadensis root for 2 years (106 weeks). At the end of the treatment, the doses of 1175 mg/kg in rats and the doses varying from 1120 to 3275 mg/kg in mice showed hepatocarcinoma [175].  Table 2. The other two plants are described in Section 4.4.1. The Naranjo algorithm [176], which consists of 10 questions that address the factors considered to determine the causal relationship in case reports, was used to assess causality. A score is obtained (maximum 13) and the results are classified as doubtful or unlikely (0), possible (from 1 to 4), probable (from 5 to 8), and clear or definite (socre > 9). The event must be definitive from a pharmacological or phenomenological point of view, using, if necessary, a conclusive procedure of reexposure [177].
Those cases that report hepatic damage were also evaluated using the method proposed by the Council for International Organizations of Medical Sciences/Roussel Uclaf Causality Assessment Method (CIOMS/RUCAM) [178], which is an organospecific instrument designed for the assessment of hepatotoxicity. This method evaluates the temporal relationship between the consumption of a substance (drug or natural remedy) and the appearance of hepatic damage, the absence or presence of risk factors, the exclusion of alternative causes of liver injury, among others. The sum of the scores leads to a final score between −8 and 14 points, which results in the following categories: highly probable or definite, probable, possible, or excluded. The amount of information of each clinical case considered for this review was classified as enough (number of criteria: 5-6), regular (number of criteria: 3-4), and poor (number of criteria: 1-2). The following criteria were used to evaluate the amount of information: (1) clear information regarding the intake and time elapsed for the onset of symptoms, (2) information of the dose ingested, (3) explanation and clinical management of the intoxication, (4) information for the exclusion of other causes that might induce the toxic effect, (5) information of the withdrawal of the plant substance, and (6) time of recovery from the toxicity or death of the patient.
The toxicity presented in clinical cases was mainly due to the accidental consumption of toxic medicinal plants, especially by children. In all the cases, the toxic effects occurred after the administration of the plant. The symptoms of toxicity were confirmed using objective evidence. None of the reports provided information about the presence of similar toxic effects compared to a previous experience. Improvement of symptoms occurred in some cases (i.e., [58, 62-64, 69, 71, 72]). The information about the number of ingestion with the plant is only reported in some cases (i.e., [43,57,59,60,65,67,70]).

Case Series.
Krenzelok et el. [179] gathered information about Euphorbia pulcherrima exposure during an 8-year period in the United States of America. The results showed that children accounted for 93.3% of cases of exposure, which were accidental (98.9% of cases) and by ingestion (94.5% of cases). No deaths were reported. However, this study did not report the symptomatology. The toxicity of Cimicifuga racemosa has been extensively studied. The reviews of Borrelli and Ernst [180] and Teschke et al. [181] can be consulted for more information regarding the adverse effects of Cimicifuga racemosa in other clinical cases. The prevalence of allergy to Myroxylon pereirae resin has been reported in many countries, ranging from 5.4 to 11.8% (i.e., [49,[182][183][184][185]). From a total of 27815 patients recorded over 5 years in Croatia, 8.4% were positive to contact dermatitis upon exposure to Myroxylon pereirae bark [186]. In another case, the prevalence of toxicity by medicinal plants was also reported. Jatropha curcas, Andira inermis, and Canella winterana were the third, the fourth, and the fifth most cited plant species, respectively, associated with cases of toxicity in Cuba from 1998-2007 [36]. Eddleston et al. [187] reported 351 patients with a history of T. peruviana consumption for 2 years. No deaths were reported.

General Considerations
The identification of the compounds responsible for the toxicity has been reported only in some cases. Urushiol might be the compound responsible for the dermatitis reactions to Metopium brownei [188], whereas sanguinarine is the compound associated with the toxicity of Argemone mexicana. Eddleston et al. [189] reported six fatalities in patients who ingested between 1 and 10 seeds of T. peruviana. These fatalities occurred due to high concentrations of cardiac glycosides (neriifolin, thevetin A, thevetin B, and oleandrin) in seeds. Three toxins (T-544, T-514, and T-496) have been reported in Karwinskia humboldtiana [190]. Manihot esculenta, an important dietary staple, is toxic because of the presence of cyanogenic compounds. Linamarin, the predominant cyanogenic glycoside in Manihot esculenta, can be accumulated in a range of concentrations between 100 and 500 mg/kg in roots and leaves. The content of HCN in Manihot esculenta has been reported 0.1-1 mg/g fresh weight in the leaves [191]. Several intoxications have been described in humans. The clinical pattern consists of neuropathy and hyperthyroidism [192]. The mechanism of toxicity is also unknown in many cases. The mechanism of toxicity of Argemone mexicana oil might be explained by the inhibitory effects of sanguinarine on Na+/K+ATPase, induction of cell membrane damage by reactive oxygen species and lipid peroxidation, and inhibition of DNA polymerase activity, among other effects [193]. Larrea tridentata and nordihydroguaiaretic acid, its active compound, generate acute hepatotoxicity by the inhibition of cyclooxygenase and cytochrome P-450 [194].
Special attention should be given in medicinal plants such as Argemone mexicana, Chenopodium ambrosioides, and Thevetia peruviana. Effects in humans have been reported due the consumption of these medicinal plants. Other plant species, including Abrus precatorius, Capsicum annum, Conium maculatum, Erythrina Americana, Heliopsis longipes, Hydrastis canadensis, Jatropha curcas, Jatropha gossypifolia, Karwinskia humboldtiana, Larrea tridentata, Magnolia grandiflora, Parthenium hysterophorus, Phoradendron serotinum, Plumeria rubra, Prosopis juliflora, Ruta chalepensis, Sanguinaria canadensis, Solanum americanum, and Veratrum californicum, have shown effects considered highly toxic, including hepatotoxicity, teratogenic, and cardiotoxicity, or with high toxicity in acute studies. Therefore, a total of 22 plants of 216 cited in this review should be extensively studied in terms of their toxicity. Regarding the hepatotoxicity induced by medicinal plants, Valdivia-Correa [39] reported 15 medicinal plants commonly used in Mexican traditional medicine that induce hepatotoxicity. Five (Equisetum hyemale, Tilia mexicana, Passiflora edulis, and Scoparia dulcis) of these fifteen medicinal are considered as toxic, according to our bibliographic research. The induction of hepatotoxicity induced by herbal products represents a serious problem in Mexico since the symptoms and signs might be confused with other diseases, and the diagnosis can be incorrect [39].
Some aspects that influence the toxicity of medicinal plants reported in this study are: (a) time of exposure, (b) misidentification of medicinal plants, and (c) adulteration of medicinal plants. Most of the acute symptoms reported in this review appear in the first 24 h after exposure. Nevertheless, more studies, including subacute and chronic assays, as well as the quantitation of hepatic enzymes, should be performed. In other cases, such as the intoxication of Crotalaria sagittalis, the toxic symptoms appear 2 to 6 months after the exposure [4]. However, chronic poisoning induced by medicinal plants is not easily detected since the symptoms are multiple and variable and a diagnosis cannot be made. Many poisonings caused by medicinal plants are classified as of unknown origin because most of the patients deny the consumption of herbal products. For instance, the clinical picture of intoxication with Karwinskia humboldtiana might be confused with poliomyelitis [190]. In addition, in most of the cases, the plants are not taxonomically identified [36]. The misidentification of medicinal plants represents a serious problem for human health. The adulteration of medicinal plants sold as food products should be considered as a risk of intoxication by medicinal plants.
Another aspect to consider for further studies is the evaluation of mixtures of medicinal plants and the combination of medicinal plants with allopathic medications. It is thought that the combination of medicinal plants might result in higher beneficial effects compared to those found with single preparations. Nevertheless, it might be the case that two toxic plants are combined and their toxic effects might result in a synergistic action. The self-medication of drugs along with the consumption of medicinal plants is a common practice among patients with chronic diseases [195], which can be considered as an alternative cause of intoxication. In the clinical record, it is not indicated whether the patient consumes medicinal plants. The interaction of herbal extracts and drugs remains to be studied. There are few documented cases that report the toxicity of the combination of plant extracts and drugs. For instance, the combination of Picrasma excelsa coumarins are reported to potentiate the activity of warfarin [196]. The toxicity of mixtures of medicinal plants and the combination of medicinal plants with allopathic medication requires further investigation.

Conclusions
There is limited information about the toxicity of medicinal plants used in Mexico, Central America, and the Caribbean. The molecular mechanisms by which medicinal plants induce toxic effects should also be addressed. In many cases, intoxication by medicinal plants might be confused with other diseases. The detection of intoxication with medicinal plants could be difficult because the symptoms might be confused with other diseases.
The prevention of poisoning in humans can be avoided if the chemical composition of medicinal plants is known.