This study was conducted in ten different sites across the DRC for monitoring molecular resistance of P. falciparum to CQ and to AQ. The study has shown that sixteen years after official CQ withdrawal from the national treatment policy, CQ-resistance prevalence has decreased in DRC to 28.5% and is marked by a high variability from site to site that was found to be significantly different (p<0.001). No molecular marker of AQ resistance has been found.
A decline in the prevalence of resistance has also been reported in several other African countries. An in vivo assay conducted in 2005  revealed a return of CQ efficacy to 99% in Malawi, the first country to change its national first-line malaria treatment policy from CQ to SP in 1993. More studies have shown the return of efficacy for CQ several years after its withdrawal from treatment policies. In Tanzania, CQ resistance decreased from 80% in 2001 to 5.7% in 2011 . In Kenya, it decreased from 76% to 6% between 2003 and 2015 . In the Republic of Congo (Brazzaville), resistance decreased from 100% in 2005, one year before the introduction and implementation of ACT in 2006, to 71% in 2015 . In Zambia, no CQ resistance marker was detected in a clinical trial conducted from 2010 to 2013 .
The relationship between drug pressure and CQ sensitivity has been clearly reported by Feng et al. and Frosch et al. [16, 29]. When the pressure stops, the drug tends to recover its effectiveness against the parasite. Mutations are common in parasites. However, the fitness of the mutant, which is its ability to survive compared to the wild-type parasite, may be altered as shown by a deficit of reproductive potential in a number of mutants that are otherwise viable . Consequently, the majority of mutant parasites will gradually disappear from the natural population, allowing the emergence of the wild-type population . This explanation could partly explain the observations of a very low prevalence of K76T mutations detected in Fungurume (1.5%), Lubumbashi (1.8%), Kalima (4.5%) and Kisangani (4.7%) in the present study. However, this prevalence was higher in other sites such as Bolenge (32%), Vanga (41.3%), Kinshasa (48.8%) and especially at Katana (89.5%). Previous studies have reported much higher K76T rates in Kinshasa in 2008 (83.8%)  and 2010 (73.2%) , and in Bolenge in 2014 (70.6%) .
The simultaneous presence of very low and high prevalence of CQ resistance could be related to different levels of CQ pressure between study sites, differing from one site to another, before and after the withdrawal of this molecule. Concerning the use of CQ in DRC, data from the DHS II in 2013–2014 revealed that in provinces where our sites are located (before territorial apportionment from 11 to 26 provinces) CQ was still in use despite its withdrawal from the national policy of malaria management in 2001 . Unfortunately, these DHS data have not yet been updated in the 26 current provinces and the low remaining use of CQ cannot fully explain the disparity of CQ resistance observed in different locations. Thus, further studies at the community level should be conducted to enrich the data.
The Katana site which had the highest rate of CQ resistance (89.5%) is located in the eastern province of Sud-Kivu, where several armed militias have been warring during the last two decades. This instability in the country’s security is an obstacle to good management of the use of anti-malarial drugs as recommended by the national malaria treatment policy. Lack of control of the anti-malarial supply chain could result in the use of non-recommended molecules by the population, such as CQ. In contrast, other sites such as Lubumbashi in the southeast of the country, where the security situation has been calmer, have seen a significant decrease in the prevalence of the K76T mutation in parasites. Another distinct feature of Katana is its epidemiological facies. The Katana site corresponds to the mountainous facies, where a lower malaria transmission rate results in a low rate of sexual recombination of parasite genotypes in the mosquito. As a consequence, the drug resistant genotypes become more established in the host population. However, Bushman et al. have recently suggested an alternative hypothesis based on the competition between drug-sensitive and drug-resistant parasites within the human host. This competition could slow the spread of resistance in high-transmission settings, which are marked by mixed parasite strain and genotype infections . Conversely, human malaria infections consisting of multiple parasite genotypes are rarely observed in low transmission settings. Drug resistance appears and emerges from this kind of place; as previously shown in Southeast Asia, considered to be the bastion of anti-malarial resistance . In Africa, anti-malarial drug resistance has historically risen in the east and spreads to the rest of the continent . As a reminder, the first case of resistance to CQ was reported in Tanzania . The province of Kivu in the eastern Congo (where Katana is located) was the first region to report CQ resistance in the DRC . In addition, a therapeutic efficacy study conducted in 2001 in the DRC, found that the city of Bukavu in Kivu contained the highest percentage (80%) of patients who had treatment failure with CQ . This finding shows a high conservation of the K76T mutation in a part of the DRC sixteen years after the discontinuance of CQ as first-line therapy in the DRC national malaria policy. Future molecular studies are necessary to monitor the trends in CQ resistance marker rates and to confirm or refute this disparity across the country.
Concerning pfcrt haplotypes, the CVIET haplotype was predominant in isolates carrying K76T mutations in this study. However, there are many other possible combinations of polymorphisms in positions 72-76 that include the key mutation K76T in CQ-resistant P. falciparum, with CVIET as the most common haplotype in Africa .. The single mutation in codon 76 is rarely observed in nature, suggesting that compensatory mutations in codon positions other than 76 could be required to restore the fitness of the CQ-resistant parasites bearing the K76T mutation . The SVMNT haplotype associated with AQ resistance was not detected in the present study which is good news for ACT use. This haplotype has not yet been reported in the DRC [20–22] whereas it was found in neighbouring countries, such as Tanzania and Angola [12, 13]. Regular monitoring of resistance to AQ is required because AQ is the partner molecule in ASAQ combination therapy, which is one of the artemisinin-based combinations currently used in the country.
Out of 806 positive P. falciparum samples sequenced, 42 samples (5.2%) have not given interpretable sequences. The yield of sequencing depended on several factors including the concentration of template used in the reaction. The discordance between screening test results (RDTs and thick blood test) and those of PCR P. falciparum detection will be addressed in a future publication.
The limitation of the study is the low sample size (not representative of all the country) that results in a partial explanation of the disparity in the CQ resistance marker rate between the study sites.