For P falciparum malaria infections during the first trimester, WHO recommends quinine with clindamycin for 7 days (or quinine alone if clindamycin is not available) and, in situations of failure or unavailability, an artemisinin-based combination therapy (ACT) or oral artesunate with clindamycin for 7 days.² This recommendation is based on data from 700 pregnant women exposed to artemisinin derivatives during the first trimester, and excludes at least a 4·2-fold increase in risk of major congenital defects.² However, the Malaria Policy Advisory Committee has recommended that these guidelines should be revised on the basis of meta-analysis findings.^21,22

Malaria caused by Plasmodium species other than P falciparum (non-falciparum malaria) should be treated with chloroquine; quinine is recommended for chloroquine-resistant infections.²

Guidelines for the treatment of P falciparum malaria in the second and third trimester are the same as for non-pregnant adults; this means any ACTs that are recommended as first-line treatment—namely, artemether-lumefantrine, artesunate-amodiaquine, artesunate-mefloquine, dihydroartemisinin-piperaquine, or artesunate plus sulfadoxine-pyrimethamine—can be used in pregnancy.² A systematic review²³ of 16 randomised controlled trials done between 1998 and 2009 presented ten trials testing ACTs (three trials of artesunate plus sulfadoxine-pyrimethamine, two of artemether-lumefantrine, three of artesunate-mefloquine, one of dihydroartemisinin-piperaquine, and one of artesunate with atovaquone-proguanil) versus either combinations without artemisinins or monotherapies. In most trials, ACTs had a PCR-adjusted efficacy of more than 90%, with the exception of artemether-lumefantrine at the Thai-Myanmar border,²³ which had an efficacy of 87% at day 42 that was attributed to low drug concentrations and low antimalarial immunity.²⁴ A systematic review and metaanalysis comparing the efficacy, safety, and tolerance of ACTs with that of quinine and other non-ACT antimalarial drugs (azithromycin plus sulfadoxine-pyrimethamine or amodiaquine plus sulfadoxine-pyrimethamine) included six trials done between 1995 and 2009; three studies were from sub-Saharan Africa (Malawi, Tanzania, and Uganda) and three were from Asia (Thailand), and all of them were included in the previous review,²³ except the study in Uganda.²⁵ ACTs were significantly more efficacious than was oral quinine in Thailand and had similar efficacy to non-ACTs in Africa. Birth outcomes were similar between treatment arms, with the exception of mean birthweight, which was significantly higher in ACT recipients than in non-ACT recipients, indicating ACTs might clear parasites (including those in the placenta) more efficiently than do other treatments.²⁶ Furthermore, artemether-lumefantrine was associated with decreased rates of moderate to high-grade haemozoin deposition in the placenta compared with oral quinine in Uganda (13–3% vs 25–8%), indicating a protective effect against placental malaria.²⁷

A large multicentre randomised open-label trial testing four ACTs (artemether-lumefantrine, artesunate-amodiaquine, artesunate-mefloquine, and dihydroartemisinin-piperaquine) in pregnant women with P falciparum malaria was done between 2010 and 2013 in four sub-Saharan African countries (Burkina Faso, Ghana, Malawi, and Zambia). In total, 3428 pregnant women were recruited and followed up until day 63 after treatment and again at delivery. PCR-adjusted cure rates for all ACTs ranged from 94·8% to 99·2%, within the prespecified equivalence margin. Nevertheless, the cure rates in the artemether-lumefantrine group were significantly lower than for the other treatments, which had similar high efficacy.²⁸ The significantly lower unadjusted cure rates in the artemether-lumefantrine group (52·5%) than in the other treatment groups (artesunate-amodiaquine 82·3%; artesunate-mefloquine 73·8%; dihydroartemisinin-piperaquine 86·9%) show that, in areas of intense transmission, dihydroartemisinin-piperaquine might be preferable to artemether-lumefantrine because of its longer post-treatment prophylactic period.

A few smaller trials done in sub-Saharan Africa (Nigeria in 2015²⁹ and Uganda in 2016³⁰) have also shown the high efficacy of ACTs.

In southern Papua, Indonesia, in 2006, dihydroartemisinin-piperaquine became the first-line treatment for malaria in pregnant women in their second and third trimester; as a result, the number of congenital malaria cases decreased from 3·2% to 0·2%, and there have been no cases since 2008.³¹ The implementation of dihydroartemisinin-piperaquine also resulted in a decreased risk of malaria at delivery, early neonatal deaths,³² severe maternal anaemia, and low birthweight.³³

Although chloroquine can be used to treat non-falciparum malaria,¹ Plasmodium vivax resistance emerged in the 1980s in New Guinea and has spread to the Indonesian archipelago and Mekong region.³⁴ Most antimalarial drugs with activity against P falciparum have intrinsic activity against the asexual stages of P vivax, except antifolate drugs.³⁴ Therefore, P vivax malaria can be treated with any ACT that is effective against P falciparum, with the exception of artesunate plus sulfadoxine-pyrimethamine.² Although ACTs rapidly clear the asexual stages of P vivax, there is high variability in the occurrence of recurrent infection between 28 days and 63 days post-treatment.³⁴ Primaquine—which is the only available drug effective against the liver stages of the parasite’s life cycle—is contraindicated in pregnant women and during breastfeeding, because of the risk of haemolysis if the offspring is glucose-6-phospate dehydrogenase deficient;³⁴ primaquine can be administered when the woman has stopped breastfeeding.

Systemic exposure to artesunate and its active metabolite, dihydroartemisinin, after oral administration of artesunate, was substantially lower in pregnant women with P falciparum malaria on the Thai–Myanmar border than in historical³⁸ and post-partum controls.^39,40 In one of these studies, malaria and pregnancy had opposite effects on the absorption of orally administered artesunate; malaria increased the oral bioavailability of artesunate by 87%, whereas pregnancy decreased the oral bioavailability by 23%.³⁹ However, there was no evidence of pregnancy-related alterations of the pharmacokinetic properties of artesunate or dihydroartemisinin after intravenous administration, suggesting that standard treatment recommendations for severe malaria apply to pregnant women. A study carried out in Kinshasa, Democratic Republic of the Congo, which compared women during pregnancy and postpartum with non-pregnant controls, showed that drug exposure changed in pregnant women (42% decrease in exposure to dihydroartemisinin) after oral administration of artesunate.^41,42 However, no difference in exposure to dihydroartemisinin was seen in pregnant and non-pregnant women in Burkina Faso after oral artesunate treatment.⁴³

Two clinical studies in pregnant women and matched non-pregnant controls in Uganda³⁰ and Tanzania⁴⁴ reported that the pharmacokinetic properties of artemether and its active metabolite, dihydroartemisinin, were unaltered after oral administration of artemether. However, studies that recruited only pregnant women showed lower drug exposures in pregnant women than did studies of historical controls.^45–47,61

Contradictory results have also been presented regarding systemic drug exposure to dihydroartemisinin after oral administration in pregnant women and matched non-pregnant controls.^48–51 In Thailand and Uganda, drug exposure was substantially lower (Thailand 38% lower, Uganda 47% lower) in pregnant women than in non-pregnant women,⁴⁹ whereas in Papua New Guinea pharmacokinetic properties in pregnant women were unaltered.⁴⁸ Therefore, it might be necessary to increase doses of ACTs for pregnant women, but more data are needed. A systematic review reached similar conclusions.⁷⁰

Drug exposure to chloroquine and its main metabolite, desethylchloroquine, was significantly reduced (chloroquine 25% reduced, desethylchloroquine 45% reduced) in pregnant women, compared with in age-matched non-pregnant women, in Papua New Guinea when receiving three daily doses (450 mg/day) of chloroquine as IPTp.⁵² This difference was due to increased elimination of both chloroquine and desethylchloroquine during pregnancy. However, in another study, pharmacokinetic parameters of chloroquine or desethylchloroquine were not different between pregnant and non-pregnant Karen women with P vivax malaria.⁵³

There were no differences in the pharmacokinetic properties of amodiaquine or desethylamodiaquine, its main metabolite, between pregnant women in the second and third trimesters with P vivax malaria and the same women at 3 months postpartum.⁵⁵ Population pharmacokinetic modelling showed that pregnancy did not have a clinically significant effect on the pharmacokinetics of amodiaquine or desethylamodiaquine, with no need for dose adjustment.⁵⁶

There have been contradictory results regarding the pharmacokinetic properties of piperaquine in pregnancy. There was no significant difference in total drug exposure to piperaquine between pregnant and nonpregnant women with P falciparum malaria in Thailand.⁵⁰ Population pharmacokinetic modelling at the population-level showed similar effects of piperaquine on the relative bioavailability and elimination, resulting in a net effect of unaltered drug exposure, but a shorter elimination half-life in pregnant women.⁴⁹ Similar results were obtained in pregnant and age-matched and weight-matched nonpregnant Sudanese women with P falciparum malaria.^57,58 However, exposure to piperaquine was about 40% lower in pregnant women than it was in non-pregnant women in Papua New Guinea⁴⁸ and Uganda.⁵¹

There were no relevant differences in exposure to mefloquine between pregnant women in their second and third trimester and matched non-pregnant women with P falciparum malaria in Burkina Faso when these women were treated with artesunate-mefloquine.⁴³ However, peak concentrations of mefloquine were significantly lower in pregnant women than in non-pregnant women with P falciparum malaria treated with a single oral dose of mefloquine.⁵⁹ Similarly, a dose-finding study on the Thai-Myanmar border suggests that drug exposure to mefloquine might be decreased in late pregnancy.⁶⁰

Mean pharmacokinetic parameters of quinine and its metabolites were not significantly different between Sudanese pregnant and non-pregnant women with P falciparum malaria who received a single dose of quinine hydrochloride (as intravenous infusion over 2 h), suggesting that no dose adjustment is required in pregnancy.⁶² However, in these women, exposure to quinine during clinical malaria was higher than it was during the convalescence phase.⁶³ Similarly, in Uganda, increased exposure to quinine during clinical malaria, compared with the convalescence phase, was reported in pregnant women with P falciparum malaria who were treated with oral quinine. Nevertheless, drug exposure in pregnant women was only about half of that in non-pregnant patients.⁶¹

Systemic drug exposure to lumefantrine is generally lower in pregnant women than in non-pregnant women treated with artemether-lumefantrine for P falciparum malaria.^44,64,65 These studies showed a decrease of about 30% in concentrations of lumefantrine on day 7 in pregnant versus non-pregnant patients. However, one study in rural Uganda showed no differences in exposure to lumefantrine between pregnant women and non-pregnant women with P falciparum malaria.³⁰

In Papua New Guinea, exposures to sulfadoxine and pyrimethamine were significantly lower in pregnant women than in non-pregnant women.⁶⁶ A study in Kenya evaluated the pharmacokinetic properties of sulfadoxine and pyrimethamine in 33 pregnant women and 11 women post partum and had similar results for sulfadoxine, while pyrimethamine was unaffected by pregnancy.⁶⁷ A multicentre study (Mali, Mozambique, Sudan, and Zambia) also showed that exposure to sulfadoxine was lower during pregnancy than it was during post partum while reporting higher pyrimethamine exposure during pregnancy.⁶⁸ Pharmacokinetic data for both drugs were highly variable among the study sites and did not suggest that dose adjustment was necessary in pregnancy.⁶⁸

In pregnant rats on gestational day 10, artemisinin derivatives have embryotoxic effects (death, cardiac malformations, and long bone malformations) due to the death of circulating embryonic erythroblasts.⁹² In human beings, dihydroartemisinin is responsible for erythrotoxicity.^93,94 In rats, embryos were most sensitive to the lethal effects of artesunate at gestational days 10–14; the corresponding gestational age in human beings is about 3–9 weeks after conception.⁹⁵ Artemisinins concentrate in infected red blood cells while malaria causes hypoferraemia.⁹⁶ Therefore, malaria might protect against artemisinin-induced decreases in the number of reticulocytes by reducing the concentrations of active drug or ferrous iron (which activates the drug to toxic free radicals), or both, in target tissues; such protection by malaria against artesunate-induced toxicity has been seen in rats. This finding could also be true for embryotoxicity, which would mean that pregnant women without malaria would be at greater risk of artemisinin-induced embryotoxicity.⁹⁵

A meta-analysis on 1664 pregnancies that were followed up after treatment with either artemisinin or quinine during the first trimester had no differences in the risk of miscarriage, stillbirth, or major congenital malformations.²² Risk of miscarriage was similar between women treated with artemisinins during the first trimester and those not treated with an antimalarial; the risk was significantly higher for women treated with quinine than for those not treated with an antimalarial drug.^22,97 In Thailand, the risk of miscarriage among women attending antenatal clinics between 1986 and 2010 was not significantly different to the risk in those treated between 6 weeks and 12 weeks of gestation with artesunate (31%), quinine (27%), or chloroquine (26%; p=0·71).⁹⁰ The risk of miscarriage associated with malaria outweighed any adverse effects from treatment with antimalarial drugs, including artemisinins.⁹⁰

In Thailand,⁷¹ pregnant women in their first trimester who were exposed to either artemisinins or quinine had a similar risk of miscarriage. Consideration of only exposure during the embryo-sensitive window (6–13 weeks gestation) showed that the occurrence of congenital malformations for artemisinins or quinine was similar, although the sample size was small (109 pregnancies for artemisinins, 641 pregnancies for quinine). In Kenya,⁹⁸ first trimester exposure to artemisinins was reported in 299 (26%) of 1134 pregnant women, in 178 (8%) of 2167 women in Tanzania,⁹⁹ in 156 (16%) of 1001 women in Zambia,¹⁰⁰ and in 96 (9%) of 1072 women in Rwanda.¹⁰¹ In Kenya,⁹⁸ the risk of miscarriage was higher among women treated with artemisinins than in women with no exposure to antimalarial drugs; however, this increase was not seen when analysis was restricted to exposure during the embryo-sensitive period, or when women treated with quinine were compared. In Tanzania,⁹⁹ adverse pregnancy outcomes (miscarriage, stillbirth, or prematurity) were more common in women treated with quinine than in women treated with any other antimalarial drug, including artemether-lumefantrine. In Zambia,¹⁰⁰ first trimester exposure to antimalarial drugs was not associated with adverse pregnancy outcomes.

A review of artemether-lumefantrine use in sub-Saharan Africa⁷² showed that receipt of this treatment in the first trimester of pregnancy did not increase the risk of perinatal or neonatal death or stillbirth. Infant neurodevelopment, birthweight, and overall incidence of birth defects were also similar, irrespective of treatment with artemether-lumefantrine or other antimalarial drugs during the first trimester. All cases of miscarriage in the artemether-lumefantrine exposure group were in patients who had received treatment during the first trimester, although in most cases there were confounding factors.⁷² Preclinical data on lumefantrine alone did not show any embryotoxicity.³⁶ Nevertheless, artemether-lumefantrine is still not recommended for the treatment of malaria during the first trimester of pregnancy unless quinine, with or without clindamycin, has failed or is unavailable.

A systematic review and meta-analysis¹⁰² of second and third trimester exposure to ACTs in studies in Africa and Asia showed that the risk of miscarriage and congenital anomalies is similar among women in the second or third trimester of pregnancy who were treated with artemisinins and women treated with quinine or other non-artemisinin antimalarial drugs. The analysis also showed that the risk of stillbirth was lower in women who were treated with ACT than in those treated with quinine, possibly reflecting a higher efficacy of artemisinin treatment.¹⁰²

Another systematic review showed that, in the second and third trimester, artemether-lumefantrine was not associated with increased adverse pregnancy outcomes as compared with quinine or sulfadoxine-pyrimethamine, showed improved tolerability relative to quinine, and its efficacy was non-inferior to quinine.⁷³

Between 1948 and 1990, six studies reported amodiaquine use in pregnancy; only one study had adverse events, but information on these events was scarce.¹⁰³ A subsequent study showed that amodiaquine, given alone or in combination with sulfadoxine-pyrimethamine during the second or third trimester, was not associated with liver toxicity or bone marrow depression.¹⁰⁴ In Ghana, pregnant women treated with amodiaquine alone or in combination with sulfadoxine-pyrimethamine had an increased frequency of mild adverse events compared with those treated with sulfadoxine-pyrimethamine alone; there was no difference in miscarriages, stillbirths, neonatal jaundice, and neonatal deaths between the groups.¹⁰⁵ At standard doses, amodiaquine does not cause developmental malformations of the embryo or fetus and the adverse events seen during pregnancy are no more common than those associated with P falciparum malaria in pregnancy.¹⁰⁶ Since this study¹⁰⁶ was published, a smaller study reported that amodiaquine is safe and reasonably well tolerated.⁵⁵ Amodiaquine-artesunate was not associated with adverse birth outcomes in a 2009 study in Tanzania.⁷⁴ Similarly, the proportion of women who reported adverse events during the 7 days after treatment did not differ significantly between treatment groups (IPTp with sulfadoxine-pyrimethamine, and treatment with sulfadoxine-pyrimethamine or amodiaquine-artesunate) with the exception of general weakness, which was slightly more common in women treated with amodiaquine-artesunate.⁷⁵

Dihydroartemisinin-piperaquine was well tolerated and had an acceptable safety profile in one arm of a trial in which more than 800 African pregnant women were treated in the second and third trimester.²⁸ These results are similar to those from other smaller studies done in Asia^50,76 and Africa.^57,58 Although dihydroartemisinin-piperaquine can cause prolongation of the QT interval,⁷⁹ no clinically significant prolongation of the QT interval was seen on 42 pregnant women receiving dihydroartemisinin-piperaquine.^77,78

Initial concerns regarding the association between mefloquine and stillbirth came from a retrospective analysis in Thailand.¹⁰⁷ This finding was not supported by earlier studies that evaluated mefloquine for treatment of malaria in pregnancy, nor by later studies on mefloquine-artesunate.^108–110 The prevalence of birth defects and fetal loss were similar to background rates in 2506 pregnant women exposed to mefloquine.⁸³ In 850 African women in the second and third trimester of pregnancy who had P falciparum malaria, mefloquine-artesunate was less well tolerated than was artemether-lumefantrine and dihydroartemisinin-piperaquine, and drug-related adverse events were more common with mefloquine-artesunate than with artemether-lumefantrine or dihydroartemisinin-piperaquine; pregnancy outcomes were similar to other antimalarial treatments.²⁸ When mefloquine alone was used as IPTp, incidence of spontaneous abortions, stillbirths, and congenital anomalies did not differ significantly from incidence in the sulfadoxine-pyrimethamine groups, although adverse events were more common.⁸⁴ Adverse events were common, but mostly minor, in a study of pregnant Beninese women; 61 (65%) of 94 HIV-positive women and 300 (78%) of 385 HIV-negative women to whom two doses of mefloquine was given as IPTp had adverse events. Notably, mefloquine tolerability was better in HIV-positive women, a finding that might be explained by these women being more familiar with adverse events and thus less prone to report them.^111,112 In two studies^113,114 for prevention of malaria in HIV-positive and HIV-negative women in Africa, mefloquine was less well tolerated than was sulfadoxine-pyrimethamine. In the study of HIV-positive women,¹¹³ the viral load and frequency of mother to child transmission of HIV was higher in the mefloquine group, but this result needs to be confirmed.

Sulfadoxine-pyrimethamine has been used extensively in pregnancy for treatment and IPTp, but formal safety studies are scarce.³⁶ Pyrimethamine causes dose-dependent embryotoxicity in rats, but not at human-equivalent doses.⁸⁷ In a case-control study, mothers whose babies had cleft palate had a higher exposure to sulfonamides than did controls.⁸⁷ Nevertheless, although use of folate antagonists in the first trimester is associated with neural tube defects, large case-control studies have shown that sulfadoxine-pyrimethamine given as IPTp does not increase the risk of teratogenesis.⁸⁵ In Malawi⁸⁰ and Sudan,⁸¹ sulfadoxine-pyrimethamine plus artesunate given to pregnant women with P falciparum malaria was safe and well tolerated, although the sample size in both countries was small. Similarly, in The Gambia, exposure to sulfadoxine-pyrimethamine and a single dose of artesunate in pregnant women during a mass drug administration programme did not have any teratogenic, or any other harmful, effects.⁸² Sulfadoxine- pyrimethamine should not be given concurrently with co-trimoxazole because of their redundant mechanisms of action and synergistic worsening of adverse drug reactions.⁸⁶ No clinical association between sulfadoxine- pyrimethamine and kernicterus has been reported.⁸⁵

The use of quinine in pregnancy is generally thought to be safe, and it is not associated with poor birth outcomes.³⁶ Quinine causes prolongation of the QT interval, but no significant cardiotoxicity has been seen in large prospective studies.⁸⁸ Furthermore, quinine can sometimes cause hypoglycaemia, particularly in the second and third trimester, even in uncomplicated malaria.⁵ In Uganda, the percentage of patients treated for uncomplicated malaria during pregnancy with at least one adverse event (most commonly tinnitus) was significantly higher in the quinine than it was in the artemether-lumefantrine arm.²⁵

Chloroquine has been described as safe throughout pregnancy.⁸⁹ Although chloroquine causes prolongation of the QT interval, no significant cardiotoxicity was reported in large prospective studies.⁸⁸ A study in Thailand showed that the risk of miscarriage was similar among women treated with chloroquine, quinine, or artesunate.⁹⁰