Historical trends and new surveillance of Plasmodium falciparum drug resistance markers in Angola

Anti-malarial drugs have long been important tools for malaria control [1]. However, their efficacy is constantly threatened by the evolution of drug resistance in Plasmodium falciparum [2]. Multiple P. falciparum genes are involved in drug resistance, and selection of them varies by allele, genetic background, and drug environment [3‐5]. Therefore, frequent monitoring of resistance alleles is crucial to predicting the spread of drug resistance. This is especially true in the West African country of Angola, where malaria cases and deaths are on the rise [6].

The first anti-malarial drug to enjoy widespread use in Angola was chloroquine (CQ) in the 1950s [7]. CQ resistance was first confirmed in Angola in the 1980s, and by the early 2000s, CQ failure rates exceeded 80% [8, 9]. As a result, CQ was discontinued in Angola in favour of artemisinin-based combination therapy (ACT) starting in 2006 [6]. To discourage the evolution of artemisinin resistance, artemisinin is used in combination with the longer-acting partner drugs lumefantrine or amodiaquine, which is chemically related to CQ [10]. Artemisinin resistance has not yet appeared in Angola, although many partially resistant kelch13 mutations have emerged in Southeast Asia [5, 11]. Nonetheless, occasional ACT failures have been reported in Angola that could be due to partner drug resistance [6, 12].

Strong P. falciparum resistance to CQ and amodiaquine is caused by crt 76T, a lysine to threonine substitution at codon 76 of the chloroquine resistance transporter (Table 1). A meta-analysis found this allele to be 7.2-fold overrepresented in CQ treatment failures [13], reflecting its selection by CQ and amodiaquine in many clinical studies (Table 1). In Angola, 76T is typically found on the haplotype crt 72–76 CVIET, which is of Asian origin [14]. CQ resistance has also evolved independently through the haplotype crt 72–76 SVMNT in South America and Papua New Guinea [15].

The 86Y allele of mdr1, or multidrug resistance gene 1, also confers resistance to CQ and amodiaquine [13]. Although this specific polymorphism dominated early studies of mdr1 and CQ resistance, the evolution of mdr1 is complicated by linkage between position 86 and other functional polymorphisms [16]. Precise mdr1 haplotypes vary among P. falciparum populations and drug settings, but in Angola alone, at least six alleles at three mdr1 positions have been proposed to modulate resistance to CQ, amodiaquine, and lumefantrine (Table 1, Additional file 1: Table S1).

The drug sulfadoxine-pyrimethamine (SP) has also been in widespread use in many African countries since the 1960s [17]. Plasmodium falciparum quickly began evolving partial resistance to SP, mediated by numerous substitutions in dhps and dhfr [18]. The risk of SP treatment failure increases with the number of mutant alleles present, with “quintuple mutants” at codons 437/540 of dhps and codons 51/59/108 of dhfr of particular concern [19‐21]. By the early 2000s, these alleles were common in Angola and 25–39% of P. falciparum infections failed to respond to SP treatment [9]. SP has since been discontinued as a first-line therapy in Angola, but it is still available at private pharmacies [22], where it comprised 10–40% of all anti-malarial sales in Huambo between 2009 and 2013 [23]. Intermittent, preventative SP is also recommended for pregnant women in Angola [24], although data from 2015 to 2016 indicate that only 30–40% of pregnant women actually received it during prenatal visits [25]. In other African countries, additional dhps mutations are emerging that appear to threaten the efficacy of SP treatment [26, 27]. It is, therefore, critical to continue monitoring genetic variation in dhfr and dhps.

In this work, 50 P. falciparum infections from Cabinda, Angola were genotyped for 13 markers of drug resistance in the genes crt, mdr1, dhps, dhfr, and kelch13. Similar data were also gathered from studies published on Angolan P. falciparum in the last two decades. For every gene but kelch13, the observed temporal patterns of allele frequency change are consistent with the current usage and availability of anti-malarial drugs. This work can inform future decisions on drug administration in Angola.

Discontinuation of CQ in several African countries including Malawi, the Gambia, Kenya, Ethiopia, Tanzania, and Grand Comore has led to major declines of crt 76T [38‐43], the most important allele for CQ resistance in P. falciparum. Six studies conducted in Angola since 2010 similarly found reduced frequencies of crt 76T compared to the early 2000s (Table 3). However, precise frequency estimates have ranged widely among studies (30–89%, Table 3), perhaps because of local differences in anti-malarial use or parasite diversity. CQ is still available in some Angolan pharmacies, although it comprised fewer than 1% of sales in Huambo between 2009 and 2013 [23]. Another important source of selection could be the ACT partner drug amodiaquine (Table 1), which is found in one of two artemisinin-based combinations currently implemented in Angola. More information on anti-malarial drug usage across the country, as well as continued genetic monitoring of P. falciparum, will be useful for understanding the evolution of crt in Angola.

Discontinuation of CQ in Grand Comore and the Gambia also preceded the decline of mdr1 86Y [38, 40], the second most important allele for CQ resistance in P. falciparum. Several studies from across Angola are consistent with a steady decline of 86Y after 2007, with the exception of a 2011 household survey in Benguela (Table 4). One possible explanation for the faster loss of mdr1 86Y, as opposed to crt 76T, could be that mdr1 N86 or its linked alleles are involved in low-level lumefantrine resistance (Table 1; Table S1). Lumefantrine is a component of artemether-lumefantrine (AL), the most common ACT currently implemented in Angola [6, 44]. Since 2013, three large studies have reported that AL treatment efficacy in Zaire or Lunda Sul fell below the WHO standard of 90% [12, 45, 46]. These treatment failures have been interpreted as signs of decreased susceptibility to lumefantrine in the parasite population [12], although the genetic basis of lumefantrine resistance has yet to be conclusively demonstrated and could involve multiple loci. Further studies on the mechanisms of lumefantrine tolerance in P. falciparum will be key to enabling molecular monitoring in the future.

In contrast to markers of CQ resistance, markers of SP resistance have been on the rise in several African countries [40, 41, 47‐49] where SP is used for malaria treatment and prevention. In Angola, the available data suggest a rapid increase in the prevalence of SP-resistant “quintuple mutants” in dhfr/dhps between 2004 (0%) [36], 2013–2016 (11.6%) [37], and 2018 (Table 2, 37.5%). This increase could be partially driven by the rise of one particular marker, dhfr 59R (Table 5), although dhfr/dhps genetic diversity may also vary among sites [50]. In Angola, it is likely that unregulated consumer use of SP [22, 23] and use of SP for intermittent preventive treatment in pregnancy [51] are sources of selection for SP resistance. This situation should continue to be monitored closely to avoid the eventual loss of important SP benefits during pregnancy [27, 52]. In the future, complete reporting of haplotype information for all combined dhfr/dhps alleles in each P. falciparum infection (e.g. Additional file 3: Table S3) will help accomplish this goal.

These interpretations of allele frequency change in Angola over time are subject to a number of limitations. First, the historical data were drawn from studies conducted in several geographical locations across Angola (Tables 3, 4, 5; Additional file 4: Table S4). Because provinces vary in patterns of malaria transmission [6] and anti-malarial availability [22, 23], selection on resistance alleles is not expected to be uniform across the entire country. Second, each study varied in its sampling and many also varied in their reporting of mixed infections (Tables 3, 4, 5; Additional file 4: Table S4), which has the potential to reduce or inflate estimates of mutant allele frequencies. Third, several studies (including this one) provide data for a relatively limited number of subjects, which may also lead to biases in allele frequency estimates. Despite these limitations, however, the apparent trends of drug resistance alleles over time (Tables 3, 4, 5) are consistent with the common usage of ACT and SP and discontinuation of CQ across Angola since the mid-2000s.

Finally, we detected no signs of alleles in kelch13 that confer partial resistance to artemisinin. This result is consistent with the lack of evidence in Africa for the spread of alleles that diminish ACT efficacy, although some countries have recently reported the presence of validated markers [53]. Continued monitoring of kelch13 in Africa is important, as artemisinin remains the cornerstone of anti-malarial drug policy in many countries.

Multiple drug resistance alleles in Angolan P. falciparum have experienced changes in frequency since CQ was officially discontinued in 2006. Markers of resistance to CQ appear to be declining, but markers of strong SP resistance and potential low-level lumefantrine resistance are rising. Continued monitoring and drug policy adjustments will be necessary in the future to regain control of P. falciparum malaria in Angola.