Time to scale up molecular surveillance for anti-malarial drug resistance in sub-saharan Africa

Molecular markers offer an additional strategy to monitor the early emergence and spread of anti-malarial drug resistance, are not impacted by host immunity, and may be more cost effective when implemented for routine surveillance. Retrospectively, it has been suggested that partial sulfadoxine-pyrimethamine resistance had multiple origins including areas of high transmission in Eastern Africa [22, 23] and, interestingly, the same region seems to be a hotspot for partial artemisinin resistance [12, 24]. Today, tools (molecular markers) to closely monitor the early emergence and spread of artemisinin and partner drug resistance are available. This knowledge should be used to establish a comprehensive molecular surveillance system to avoid the mistakes from the past, when the spread of resistance to anti-malarial monotherapies has been detected at a late stage, contributing to thousands of deaths in the meantime. Molecular markers cannot predict treatment outcome at an individual level, however their increase often precedes that of treatment failures [25]. While monitoring Pfk13 mutations is of paramount importance, monitoring partner drug resistance molecular markers is also crucial [26]. Indeed, high prevalence of Pfk13 validated molecular markers is not usually associated with treatment failure [27], as evidenced by recent data from Rwanda where the efficacy of both first and second-line treatments is still high despite the increasing prevalence of the Pfk13 561 H mutation (Table 1). Currently, molecular surveillance is often done retrospectively based on convenience sampling, and does not provide an accurate estimation of resistance on a national or sub-regional level. For example, when looking at available molecular data for Rwanda, there are large spatiotemporal gaps [10‐12, 28], and no clear trend is discernible (Table 1). However, high prevalence of the confirmed artemisinin resistance marker 561 H at two sites (Masaka and Rukara) are worrisome, especially as samples have been collected in 2018, and the current situation could be worse. Moreover, the prevalence of this marker has increased from 7 to 20% in Masaka between 2015 and 2018 (Table 1), even though the small sample size does not allow for definitive conclusions, but it is likely that the prevalence is even higher now, and only routine molecular monitoring could accurately assess the trend.

Molecular routine surveillance should not only be conducted in a few sentinel sites, but rather on a large network of health facilities to capture the complex spatial dynamics of evolving resistance [29]. Samples collected from patients attending selected health facilities in the different regions of a country, should be analysed on a regular basis and generated data used to map the spatiotemporal dynamics of molecular markers of interest [29‐32]. The data could be then used to inform and calibrate mathematical models of malaria transmission aiming at guiding interventions strategies, for example to select the sites for TES, using specific thresholds for artemisinin and partner drug resistance markers prevalence, to assess either delayed parasite clearance or treatment failure, respectively [21, 33]. Indeed, it is difficult to predict where resistance will emerge, and a more flexible scheme with rotating sites for TES based on molecular markers prevalence and models prediction may be more appropriate for early detection of resistance. Mathematical models could be used as well to predict when treatment failure could occur based on the molecular markers prevalence, giving more time for policymakers to prepare the change in anti-malarial drug treatment policy [34, 35].

Logistically, sample collection would require only dried blood spots collection at selected health centres when patients have a confirmed malaria diagnosis. The molecular analysis could be centralized at regional, national or sub-regional laboratory to maximize the cost effectiveness of the surveillance system [36]. With the increasing availability of high throughput techniques and assays for molecular markers of resistance genotyping in malaria endemic countries, there is an opportunity to strengthen the capacity of National Malaria Control Programmes (NMCPs) for molecular monitoring of anti-malarial drug resistance [37, 38]. Cross-border collaboration is also critical, especially in regions such as the Great Lakes (Fig. 1), where validated resistance markers have been detected in different countries, including Rwanda, Democratic Republic of the Congo, Kenya, and Tanzania (Fig. 1). Regional monitoring is key to track resistance, as parasites will spread quickly from one country to another. The establishment of regional reference laboratories associated with regional data repositories could facilitate the prompt detection of resistance and early implementation of mitigation strategies; and the malaria community must leverage on the different initiatives on the continent to improve access to the infrastructure and technical expertise for high throughput molecular analyses [39].

Ex vivo assays have played an important role in monitoring artemisinin resistance in the GMS [3, 15, 40, 41]. Even though they are often used for phenotypic assays to validate molecular markers of resistance, where available they can be useful to monitor resistance. Indeed, as for molecular markers, immunity is not a confounding factor for ex vivo assays, even though parasite culture may reduce the complexity of the infection by preferentially selecting specific clones. However, ex vivo assays can be a valuable tool for drugs with no validated molecular markers, such as lumefantrine and pyronaridine, the former being the partner drug of the most widely used ACT, and the latter, the most recent ACT partner drug approved by the WHO. Moreover, compared to molecular markers, ex vivo assays are difficult to implement, as fresh blood is required for parasite culture and the high intra and inter-assays variability limits their ability for spatiotemporal dynamics assessment [42].