Safety of artemisinins in first trimester of prospectively followed pregnancies: an observational study

Abstract

Background: Artemisinins, the most effective antimalarials available, are not recommended for falciparum malaria during the first trimester of pregnancy because of safety concerns. Therefore, quinine is used despite its poor effectiveness. Assessing artemisinin safety requires weighing the risks of malaria and its treatment. We aimed to assess the effect of first-trimester malaria and artemisinin treatment on miscarriage and major congenital malformations.

Methods: In this observational study, we assessed data from antenatal clinics on the Thai-Myanmar border between Jan 1, 1994, and Dec 31, 2013. We included women who presented to antenatal clinics during their first trimester with a viable fetus. Women were screened for malaria, and data on malaria, antimalarial treatment, and birth outcomes were collected. The relationship between artemisinin treatments (artesunate, dihydroartemisinin, or artemether) and miscarriage or malformation was assessed using Cox regression with left-truncation and time-varying exposures.

Findings: Of 55 636 pregnancies registered between 1994 and 2013, 25 485 pregnancies were analysed for first-trimester malaria and miscarriage, in which 2558 (10%) had first-trimester malaria. The hazard of miscarriage increased 1·61-fold after an initial first-trimester falciparum episode (95% CI 1·32-1·97; p<0·0001), 3·24-fold following falciparum recurrence (2·24-4·68; p<0·0001), and 2·44-fold (1·01-5·88; p=0·0473) following recurrent symptomatic vivax malaria. No difference was noted in miscarriage in first-line falciparum treatments with artemisinin (n=183) versus quinine (n=842; HR 0·78 [95% CI 0·45-1·34]; p=0·3645) or in risk of major congenital malformations (two [2%] of 109 [95% CI 0·22-6·47] versus eight (1%) of 641 [0·54-2·44], respectively).

Interpretation: First-trimester falciparum and vivax malaria both increase the risk of miscarriage. We noted no evidence of an increased risk of miscarriage or of major congenital malformations associated with first-line treatment with an artemisinin derivative compared with quinine. In view of the low efficacy of quinine and wide availability of highly effective artemisinin-based combination therapies, it is time to reconsider first-trimester antimalarial treatment recommendations.

Funding: The Wellcome Trust and The Bill & Melinda Gates Foundation.

Figure 1

Study profile P vivax=Plasmodium vivax. P malariae=Plasmodium malariae. P ovale=Plasmodium ovale. P falciparum=Plasmodium falciparum.

Figure 2

Frequency of first-line quinine and artemisinin treatments in first trimester and rates of falciparum malaria during pregnancy and miscarriage over time The increase in the rate of falciparum malaria in 1998 was due to the establishment of Shoklo Malaria Research Unit antenatal clinics in migrant communities.

Figure 3

Association between initial and recurrent first-trimester malaria and miscarriage Models include women lost to follow-up before 28 weeks (until gestation time last seen), but percentage calculations for delivered or miscarried do not. Models for falciparum malaria (1–4) include women that might have also had first-trimester vivax, malariae, or ovale malaria. See appendix p 8 for associations in women with only first-trimester falciparum malaria. Models for vivax malaria (5–6) exclude women who also had first-trimester falciparum malaria. Models 2 and 5 exclude women with symptomatic malaria. Models 3 and 6 exclude women with asymptomatic infections. Model 4 excludes women with uncomplicated infections. Models were adjusted for year (by stratification due to non-proportional hazards [p<0·001]), gravidity, current smoking status, and non-malaria febrile morbidity in first trimester. Age and previous miscarriage were omitted from multivariable models due to collinearity with gravidity. Adjusted results for gravidity, current smoking status, and febrile morbidity in first trimester are shown from Model 1. 146 women had recurrent first-trimester falciparum malaria (136 had two and ten had three episodes). 13 women had recurrent (two) asymptomatic first-trimester falciparum malaria. 81 women had recurrent symptomatic first-trimester falciparum malaria (75 had two, and six had three episodes). 17 women had recurrent (two episodes) asymptomatic fi0rst-trimester vivax malaria, and none miscarried. 38 women had recurrent (two) symptomatic first-trimester vivax malaria. See appendix p 8 for a table version of this figure, including univariable associations.

Figure 4

Association between first-line treatment of first-trimester falciparum malaria and miscarriage (n=1179) Models were adjusted for severity of the first falciparum malaria episode (asymptomatic, symptomatic, or hyperparasitaemic or severe), non-malaria febrile morbidity in the first trimester, and year of first consultation. See appendix p 9 for a table version of this figure, including univariable associations. *Hazard ratios are shown for treatments occurring before (<6 weeks’ gestation) and during (≥6 and <13 weeks’ gestation) the embryosensitive window. †Artemisinin rescue after quinine failure refers to artemisinin-based treatment in first trimester following failure of first-line treatment with quinine or quinine plus clindamycin. No miscarriages occurred in women who received artemisinin treatment after the embryo-sensitive window (≥13 and <14 weeks’ gestation). We did a subgroup analysis excluding women with asymptomatic malaria (n=919); the association between artemisinin treatment and miscarriage changed by <5% (appendix p 9). We did a subgroup analysis in women attending before 2007 whose gestational age was estimated from ultrasound biometry, because the accuracy of gestational age estimates affects the accuracy of the gestation time of antimalarial treatment, and quinine plus clindamycin succeeded quinine monotherapy in 2007 (n=469); associations were in the same direction but of greater magnitude (appendix p 9).