Background
Plasmodium parasites are the causative agents of malaria, and are transmitted to humans through the bites of female
Anopheles mosquitoes. Although the disease is both preventable and curable, malaria remains a significant global public health problem. Globally, there were an estimated 241 million malaria cases in 2020, 6% more than in 2019, and deaths increased by 12% to about 627,000 deaths, mainly due to service disruptions during the COVID-19 pandemic [
1,
2]. At the beginning of the twenty-first century, the large-scale implementation of preventive strategies such as long-lasting insecticidal nets and indoor residual spraying led to a decline in malaria incidence. However, the emergence of resistance to insecticides and treatments in mosquitoes and parasites, respectively, is problematic and hampers malaria vector control interventions. Therefore, additional approaches are needed to reduce the global malaria burden. Following the World Health Organization Global Technical Strategy for Malaria 2015–2030, a roadmap was drafted by global health experts for the development of ivermectin as a potential complementary malaria vector control tool [
3].
Ivermectin is a broad-spectrum antiparasitic drug, used in a wide range of infestations such as helminths, scabies, and mites. It has been licensed for human use for more than thirty years. Its safety profile has been thoroughly assessed in over 70 trials and extensive post-marketing surveillance. Billions of doses have been distributed worldwide for the control of neglected tropical diseases with a good safety profile, with the important exception of its use in
Loa loa co-endemic regions [
3‐
8].
After a mosquito has ingested ivermectin with a blood meal from a treated subject, ivermectin binds to the glutamate-gated chloride ion channels, leading to hyperpolarisation of the neuronal membrane, paralysing and potentially even killing the insect [
9]. Numerous studies have reported the lethal effect of ivermectin in different mosquito species, such as
Anopheles gambiae, Anopheles albimanus,
Anopheles arabiensis and
Anopheles stephensi [
3,
4,
10‐
15]. Exposure to ivermectin concentrations in the low ng/mL range decreases the survival and fertility of mosquitoes. The concentration that kills 50% of mosquitoes (LC
50) within 7-days ranges from 3 to 55 ng/mL in
Anopheles spp., and 178–187 ng/mL in
Aedes aegypti [
16‐
19].
Approximately 4 h after a single oral dose of 150 µg/kg ivermectin, the peak concentration is around 40 ng/mL [
20,
21]. After giving a single oral dose of ivermectin to humans, the drug levels in human capillary blood, from where mosquitoes feed, can be maintained well above LC
50 for a considerable time [
4,
11,
22‐
24]. In addition, the mean elimination half-life of ivermectin ranges from 25 to 80 h, and effective concentrations are thus maintained for several days post-administration [
21,
25]. Ivermectin, with its sound safety profile in humans, unique mode of delivery, novel mode of action, and long plasma half-life, is a promising candidate as a first-in-class malaria vector control tool. Currently, several clinical trials are investigating the effect of ivermectin mass drug administration on malaria transmission by diminishing the mosquito population in endemic regions [
3,
26‐
29]. The community delivery of ivermectin could address residual transmission, which is defined as a sustained transmission even after reaching an appropriate coverage with standard vector control tools. Residual transmission is mostly driven by mosquitoes biting outdoors, early in the evening or both, and feeding on livestock as an ecological niche [
30,
31]. A community delivery to both humans and livestock would likely increase MDA efficacy, as mosquitoes feeding on treated animals would also be exposed to ivermectin. This is considered in the design of current MDA campaigns trials such as BOHEMIA [
32].
Anopheles stephensi is expanding its geographic range worldwide and has been implicated in outbreaks of urban malaria [
33]. The World Health Organization (WHO) published a vector alert calling for active mosquito surveillance in Ethiopia and Sudan after an unusual outbreak of urban malaria. In December 2022,
An. stephensi was detected in Kenya for the first time [
34].
Interestingly, the blood of treated individuals reduces the survival of the mosquitoes beyond what is expected from plasma pharmacokinetics of the parent compound, even up to 28 days post-dosing [
11], which is longer than expected considering the estimated half-life of ivermectin (1–3 days) [
25]. As ivermectin is highly lipophilic and protein-bound (> 90%) [
35,
36], its pharmacokinetic patterns can differ according to several factors, such as sex, body mass index and feeding state, with a higher body fat percentage providing larger peripheral volume of distribution in female subjects, and concurrent food intake affecting gastro-intestinal solubility and, thereby, absorption [
37]. Due to its lipid solubility, ivermectin potentially accumulates in fatty tissue that act as reservoir, and is released very slowly over a long time period [
36,
38]. In populations with a high prevalence of malnutrition, the high protein binding would result in higher concentrations of free ivermectin, resulting in an increased drug effect and a higher risk of toxicity. However, so far it is not known whether metabolites contribute to the activity and are responsible for the observed “post-ivermectin” effect [
4,
11,
39,
40]. Here, mosquitocidal activity is seen even when concentrations of ivermectin drop below relevant LC
50 values, for example, even at 28 days post-dosing in the IVERMAL trial [
11]. This is further supported by mosquito feeding experiments showing that blood meals from human treated with ivermectin have greater mosquitocidal activity than those spiked with pure ivermectin at similar concentrations [
40].
Zeng et al. identified the structure of nine ivermectin metabolites in the presence of human liver microsomes, namely 3″-O-desmethyl-H
2B
1a (M1), 4-hydroxy-H
2B
1a (M2), 26-hydroxy-H
2B
1a (M3), 3″-O-desmethyl, 4-hydroxy-H
2B
1a (M4), 24-hydroxy-H
2B
1a monosaccharide (M5), 3″-O-desmethyl, 26-hydroxy-H
2B
1a (M6), 26-hydroxy-H
2B
1a monosaccharide (M7), 4, 26-dihydroxy- H
2B
1a (M8), and 24-hydroxy-H
2B
1a (M9) [
41]. Cytochrome P
450 3A4 (CYP3A4) is the predominant isoform responsible for the metabolism of ivermectin by human liver microsomes. In brief, ivermectin can be O-demethylated at the disaccharide moiety, undergo deglycosylation, and can be hydroxylated at the aglycone portion. The hydroxylation takes place at the hexahydrobenzofuran, the spiroketal portion of the molecule or both. Tipthara et al. confirmed these results and identified four additional metabolites including ketone and carboxy formation [
42].
This study investigated whether metabolites may contribute to the activity of ivermectin against An. stephensi. Since no reference standards for ivermectin metabolites were available, a method was set up to produce and purify nine different metabolites. Furthermore, a screening assay was established to estimate whether the metabolites are active against An. stephensi at levels observed in humans.
First, an in vitro system was developed using recombinant CYP3A4 and 3A5 isoforms to produce ivermectin metabolites, which were then purified and enriched by semi-preparative high-pressure liquid chromatography (HPLC). Secondly, we studied the pharmacokinetic properties of those metabolites in healthy volunteers who received a single oral dose of 12 mg ivermectin [
25]. Finally, blank human blood was spiked with each metabolite fraction at levels matching the maximal signal intensity observed in blood of the pharmacokinetic (PK) study participants. These samples were fed to
An. stephensi mosquitoes to assess the mosquitocidal activity over 72 h.
Discussion
The clinical applications of ivermectin are remarkably broad. This widely used drug can treat not only veterinary infections but also human ones, caused by various endo- and ectoparasites, as well as rosacea skin conditions [
46]. Nonetheless, the pharmacokinetics and therapeutic role of ivermectin metabolites has not yet been elucidated.
This study set out to develop a protocol to produce and isolate nine different ivermectin metabolites. It also shows that CYP3A4 and CYP3A5 metabolise ivermectin, while CYP3A4 readily produces all metabolites except M9 (Hydroxy-H
2B
1a)—that is mainly formed by CYP3A5. All metabolites apart from M9 are measurable in human blood after a single oral dose of 12 mg ivermectin. Pharmacokinetic analysis reveals that the mean residence time of the metabolites is shorter (MRT
last range: 13.9–33.8 h) as reported for ivermectin after a single dose of 150 µg/kg (89.5 h) [
47]. Finally, this study demonstrates that ivermectin metabolites M1 and M2 are mosquitocidal at concentrations observed in humans treated with a regular dose of ivermectin, and may contribute to the pharmacological effect of ivermectin treatments.
Zeng et al. have previously reported on the metabolic fate of ivermectin in human in vitro systems, and recently Tipthara et al. published in vitro as well as in vivo data [
41,
42]. The present investigation employed a multiple reaction monitoring LC–MS/MS method for nine ivermectin metabolites, derived from the published mass fragmentation data of Zeng et al. This study supports evidence from previous observations made by Zeng et al. that ivermectin is readily metabolised by CYP3A enzymes—as the metabolites were formed in the presence of human recombinant CYP3A4 and CYP3A5—and that the formation could be inhibited by ketoconazole, a selective CYP3A4/5 inhibitor [
41,
48]. In addition, CYP2C8 is involved in the hydroxylation of the spiroketal moiety of ivermectin [
42]. This current study found that CYP3A5 is involved in the O-demethylation of ivermectin (Fig.
1), as previously described by Tipthara et al. Moreover, another important finding is that CYP3A5 is mainly responsible for the formation of M9 (Hydroxy-H
2B
1a) and forms the metabolites M3 and M6—yet to a lesser extent than CYP3A4. The involvement of CYP3A5 in metabolising ivermectin might be of clinical relevance considering that its expression greatly varies between individuals according to their ethnicity and geographical location because of genetic polymorphisms in CYP3A4 and/or CYP3A5 genes [
49]. Possibly, pharmacokinetics, treatment response and appearance of adverse drug events might be linked to the patients CYP3A5 genotype. Importantly, single nucleotide polymorphism (SNP) in the CYP3A5 gene alter the expression levels of CYP3A5 enzymes leading to inter-individual differences in the enzyme activity. Racial differences might affect metabolism of and response to ivermectin considering that CYP3A5 is more frequently seen in Africans than Caucasians [
49]. Since neither the activity nor the toxicity of ivermectin metabolites have been investigated, the clinical significance of CYP3A5 polymorphisms and drug–drug interactions remain unclear. In light of the greater contribution of CYP3A4 to ivermectin metabolism rather than CYP3A5, it seems unlikely that pharmacogenetic differences between malaria endemic regions would be of importance.
The results of this study show that the most intense signals in the in vitro metabolism assays were recorded for M1 (desmethyl-H
2B
1a) and M2 (hydroxy-H
2B
1a). However, this does not necessarily indicate that those two metabolites are the most abundant. Yet, assays with radiolabeled ivermectin imply that M1 and M2 are the major in vitro metabolites of ivermectin [
41]. Interestingly, M1 was also produced by rat, pig, sheep and dog microsomes, which suggests that the O-demethylation of ivermectin may be also relevant for the treatment of livestock and domestic animals [
50‐
53]. Drug–drug interaction studies were carried out in animals to assess the effect of CYP inhibition and induction on the pharmacokinetics of ivermectin. Co-administration of ivermectin and ketoconazole increased exposure to ivermectin in sheep, but did not reduce the levels of metabolite M1 [
54]. The authors concluded that the observed interaction is rather due to inhibition of P-glycoprotein efflux transporters than of CYPs. The same conclusion was drawn by Hugnet et al., since ketoconazole did not decrease M1 levels in dogs but significantly increased exposure to ivermectin [
53]. On the contrary, ketoconazole did not alter the PK of ivermectin in invertebrates, most likely because it is rapidly excreted by
Aedes aegypti mosquitoes [
16]. Finally, the CYP activity in rats was increased by administering either rifampicin or phenobarbital daily for 1 week, but still the disposition kinetics was only modified for ivermectin—and not the investigated metabolites [
55]. These examples imply that the mechanism of interaction with ivermectin is rather due to interference with P-glycoprotein efflux transporters than with CYPs. Nonetheless, in this study ketoconazole was found to clearly inhibit the metabolism of ivermectin in vitro, and drug–drug interaction studies in humans are needed to assess the impact of CYP inhibition on the response and safety of ivermectin treatments. This is in line with previous studies that showed oral bioavailability of ivermectin being significantly altered in mammals when co-administered with drugs interacting with CYP3A and P-gp, two systems that are frequently co-located and act in synergy because of overlapping substrate affinities [
56,
57]. As such they do not only affect overall metabolism and distribution, leading to appreciable changes in metabolite profiles, but also pre-systemic metabolism. Here, CYP3A can reduce concentrations of substrates in gastrointestinal tissue and allow for more efficient efflux via P-gp. Pharmacoenhancement and boosting of ivermectin action and metabolism is, therefore, possible on several levels.
The LC–MS/MS method presented here was able to assess the kinetic disposition of eight ivermectin metabolites over a period of up to 72 h post-treatment in human whole blood. A previous study also detected the most abundant metabolites namely desmethyl-H
2B
1a (M1), hydroxy-H
2B
1a (M2), and desmethyl, hydroxy-H
2B
1a (M4) in human blood samples collected 24 h post-treatment [
42]. Only traces of M9 were observed in pharmacokinetic samples, either because the method is not sensitive enough, or the study participants were not expressing sufficient amounts of CYP3A5—as it is frequently observed in Caucasians [
58]. Nevertheless, the CYP3A5 polymorphism of the study participants was not assessed, which would have been required to confirm this hypothesis.
The drug concentration in the blood and the time the drug remains in circulation are important factors affecting the pharmacological effect. The results of the clinical pharmacokinetic study in humans indicates that ivermectin metabolites T
max (5–7 h) are shifted in time and occur later than ivermectin (4.4 h). Moreover, some metabolites elimination half-lives are longer (15.9–57.5 h) than ivermectin (38.9 h), allowing for an overall longer timeframe when mosquitoes can potentially be killed by certain ivermectin metabolites (M4, M1) than what would be expected from ivermectin alone. As another measure for the length of exposure to a substance, the mean residence times (MRT) of ivermectin metabolites (M1–M6) were calculated by non-compartmental analysis (NCA) of pharmacokinetic profiles (Table
1). The MRT is the average time a molecule resides in the body and reflects absorption and elimination rates. MRT was used as a measure for elimination times for both ivermectin and its metabolites, as this allows comparing the duration of exposure. After a single oral 12 mg ivermectin dose (corresponding to a mean dose of 181 µg/kg), the mean residence time (MRT
last) is shorter for some of the metabolites (MRT
last range: 13.9–33.8 h) compared to ivermectin. A reported MRT of 89.5 h was measured for ivermectin after a single dose of 150 µg/kg ivermectin [
47]. Because of limited sampling time the MRT of the compounds may be biased. In this proof-of-concept study with blood concentration measurements until 72 h post-dose of ivermectin, sampling times were chosen based on operational feasibility and empirical reasoning.
Kobylinski et al. demonstrated first evidence that metabolites might contribute to the activity of ivermectin as they observed that ivermectin spiked blood was less mosquitocidal than blood from treated humans with matching ivermectin levels [
40]. In support of this idea the screening assays showed that the metabolites M1, and M2 were active against
An. stephensi mosquitoes. In addition, M4 and M6 exhibited minor mosquitocidal properties either because the systemic exposure was lower or the metabolites are less mosquitocidal. M4 underwent demethylation on the disaccharide and hydroxylation on the hexahydrobenzofuran moiety like M1 and M2, respectively. All metabolites, which were hydroxylated on the spiroketal portion deglycosylated or both, were not or only slightly active (M6) at concentrations observed in clinical pharmacokinetic samples. However, the results presented here do not exclude that those metabolites are active at increased concentrations. In this study, a 3-day-LC
50 of about 3 ng/mL was measured for ivermectin, which is slightly lower compared to observations made by Dreyer et al., with a 4-day-LC
50 of 7 ng/mL against
An. stephensi [
12]. The LC
50 value varies greatly between
Anopheles species, insectary conditions, and with the feeding method or observation period. In
An. gambiae, it was reported that the 7-day-LC
50 of ivermectin is 3.4 ng/mL with blood fed from treated humans, and 15.9 ng/mL from spiked ivermectin experiments (in vitro mixture) [
3]. In addition,
An. stephensi were more susceptible to ivermectin in comparison to assays done with other
Anopheles species [
3]. Nonetheless, results were largely comparable to previous studies considering that the assay settings and conditions were not standardised (e.g. mosquito age, species, assay observation period). Further research is, however, needed to explore ivermectin metabolites impact on survival of other
Anopheles species.
The current study estimated that M1 and M2 stay on average 69 h and 34 h above their LC
50 value, and thus contribute significantly to the overall mosquito-lethal effect of ivermectin. This time above the lethality target can be considered as a “mosquitocidal window” of ivermectin, M1 and M2. The mosquitocidal window of ivermectin after a single oral 150 µg/kg dose for a vector with a LC
50 of 6 ng/mL was predicted to be of 55 h and 7 h for a vector with a LC
50 of 25 ng/mL [
4]. Several pharmacological strategies can increase the duration of time above the LC
50 and thus ivermectin’s efficacy in killing mosquitoes such as increasing the dose of ivermectin, employing repeated dosing regimens, and using a long-lasting drug formulation) [
59]. Since M1 was also detected in livestock (e.g. pigs, goats, and sheep), the findings reported here might also be of relevance for livestock treatment for malaria vector control and other veterinary applications [
50,
54]. Studies against nematode worms and other parasites are needed to assess the veterinary importance of ivermectin metabolites.
The present study has three evident limitations. First, the isolated fractions were analysed by multiple reaction monitoring, which was needed to reach the sensitivity to measure low-abundant metabolites. However, this detection mode is very selective and may not have recorded co-eluting constituents of the different fractions. Consequently, the overall purity of the isolated fractions could not be determined and it cannot be excluded that undetected metabolites or byproducts of the bioassay may have caused the observed mosquitocidal effects. However, the preparation of nine different blank fractions would have been very laborious, complicated considerably by the execution of the activity assays and would have substantially increased the required amount of consumables. Moreover, most of the fractions used for the treatments were not entirely pure and contained residues of the previous fraction (Additional file
1: Fig. S1 and Table S4). The screening assay showed that the fractions 1–6 were not or only slightly active (M6), so further purification of these fractions was not considered necessary. Fraction 8 contained mainly M2 but also residues of M4, whereas fraction 7 contained no M2. Consequently, the mosquitocidal effect of fraction 8 cannot be solely attributed to M2. However, M2 appears to be more active than M4, since the mosquito mortality of fraction 8 was much more pronounced than of fraction 7. Fraction 9 was reasonably pure and contained mainly M1, therefore, its activity can mostly be attributed to M1. Lastly, the metabolites were detected based on the parent mass and fragment and no elaborate structure identification was performed by e.g. nuclear magnetic resonance (NMR) or high-resolution mass spectrometry. Therefore, structure confirmation of at least the most active metabolites M1 and M2 by adequate analytical methods or synthesis of the reference substances is required to substantiate the findings of this present study.
Secondly, we did not investigate phase II metabolism of ivermectin, even though conjugation with glucuronic and sulphuric acids has been identified in sheep [
52]. The pharmacological and toxicological role of phase II ivermectin metabolites remains unknown. In addition, Tipthara et al. identified further ivermectin metabolites, namely M10 and M12 detected in vitro after incubation with CYP3A4 microsomes and M13 with CYP2C8. However, these metabolites were not detectable in hepatocytes or blood. Hence, it may be argued whether to assess or not the mosquitocidal activity of these metabolites in further work.
Finally, the pharmacological activity of the metabolites was investigated using levels monitored in humans after the application of a therapeutic ivermectin dose. The tested metabolite concentration had to be matched with C
max levels obtained from pharmacokinetic profiles because no references were available. Hence, the applied metabolite quantity used to treat the mosquitoes is unknown. It cannot be ruled out that metabolites judged as inactive in the presented setup might be active at higher levels and possibly also more potent than ivermectin itself. In addition, the bioassay recorded the effect over three days, which is shorter compared to assays employed by others [
11]. It is possible that more pronounced effects would have been detected for e.g. M4 and M6 by prolonging the duration of the assay. Still, the findings suggest that mainly M1 and M2 reach pharmacologically relevant concentrations in vivo and contribute to the mosquitocidal activity of ivermectin treatments.
Nonetheless, if metabolite activity or pharmacokinetic properties such as extended effect—not only against arthropods, but also in other indications—prove to be clinically relevant, future studies could explore the use of co-medications that interact with the metabolism of ivermectin. For instance, co-administering inhibitors of CYP3A may boost exposure to the parent drug, whereas inducers of the same pathway could increase production of active metabolites. Less toxic inhibitors of CYP3A and P-gp, for instance low dose ritonavir or cobicistat, should be considered.