The genetic drug resistance profile was established in pre-treatment samples from malaria patients in Karimui and the Wosera by the use of a new DNA microarray-based technology  and its relationship with in vivo drug response to the combinations of AQ+SP was analysed. The principal objectives were to establish the baseline prevalence of polymorphisms in genes related to AQ/CQ and SP resistance, to assess their relationship with treatment outcome, in order to identify and propose useful markers for molecular monitoring of drug resistant P. falciparum in the country.
The analysis of the genetic profile of the parasite population revealed high levels of mutant alleles in CQ resistance (CQR) related pfcrt and pfmdr1 genes. The long history of 4-aminoquinoline use as monotherapy in PNG has led to a highly CQ resistant genetic background in the parasites as reported previously [21, 28]. In addition, the results demonstrated prevalence rates of 91% and 82% for mutant alleles in the pyrimethamine related gene loci pfdhfr S108N and C59R. Mita et al.  recently analysed P. falciparum isolates from patients attending town clinics in Wewak (East Sepik Province) and observed similarly high prevalence rates of pfdhfr double S108N+C59R mutations (83% in 2002 and 86% in 2003). These high levels of mutation rates in pfdhfr appearing only a short time after the implementation of SP as one component of the official first-line policy were not surprising and may be due to i) the increasing recourse to SP as second-line therapy with quinine in the late 1990s (Nsanzabana et al., unpublished), ii) the former drug pressure exerted by the use of pyrimethamine (in combination with CQ) in mass drug administration campaigns in the 1960s and 1970s , and iii) the widespread use of trimethoprim-sulphamethoxazole for the treatment of bacterial infections [31, 32]. Recent microsatellite analysis in dhfr -flanking regions by Mita et al.  revealed that the most prevalent dhfr haplotype (i.e., S108N+C59R double mutation) was associated with reduced microsatellite variability around the gene, an observation which argues for the selection of pre-existing SP resistant parasites, rather than the frequent emergence of de novo mutations in this gene . These data further corroborate the hypothesis, that former drug pressure has lead to the emergence of pyrimethamine resistant parasites before the official introduction of SP in PNG.
Until 2003, polymorphic pfdhps loci associated with reduced sensitivity to sulpha drugs have only been found in a single P. falciparum isolate originating from PNG [21, 22]. In the present study, prevalence rates of 13% for A437G and 1% for K540E were observed. Likewise, Mita and colleagues detected mutations in these loci in 8% of patient isolates collected in Wewak in the year 2003. In the view that pfdhfr mutations usually predominate over those in pfdhp [12, 15], the detection of genotypes having a single dhps A437G mutation in combination with pfdhfr wild type alleles in two of the samples was rather unusual. However, this genotype may well have been selected by sulpha drugs used to treat infectious diseases other than malaria.
In order to propose a suitable marker set for the molecular monitoring of P. falciparum against the current combination therapy, the association of single mutations as well as infecting genotypes with in vivo treatment response was investigated. Regarding CQ relevant markers, the only single marker associated with a significantly increased risk of treatment failure was pfmdr1 N86Y. Taking into account additional SNPs in pfcrt, neither of the mutated alleles increased the predictive value for pfmdr1 N86Y, the most likely reason being that these mutations nearly reached fixed levels in the parasite population. Similarly, pyrimethamine relevant markers in pfdhfr did not show a significant association with treatment failure. Risk of failure was only increased with infections harbouring the A437G mutation in pfdhps. These observations are in agreement with previous studies showing that the prevalence of single molecular markers (e.g. pfcrt K76T or pfdhfr S108N) was almost always higher than the level of clinical or parasitological resistance to the respective drugs, especially in regions with high transmission intensity and long lasting drug pressure [4, 34] and therefore, renders these markers unsuitable for molecular monitoring. Furthermore, the validity of molecular markers is dependent on former drug use and may also vary according to the malaria epidemiology in a given area [10, 17, 18]. The evaluation and assessment of a combination of markers, instead of single markers indicating the presence of a highly resistant genotype, have been suggested for the molecular monitoring of antimalarial resistance [12, 14–16, 35, 36]. In the present study, which took into account the combined pfcrt/pfmdr1/pfdhfr/pfdhps genotype, the risk of treatment failure was clearly associated with the total number of mutations in the analysed genes. The risk was significantly increased for patients harbouring parasites with the most highly mutated genotype (i.e., 8/24 SNPs mutated). However, unusual findings included the increased risk of treatment failure with genotypes having the N86Y mutation in pfmdr1 and the quadruple mutation in pfcrt combined with a fully wild type pfdhfr + pfdhps allele. These results highlight again the fact that among many parasite and host factors, the molecular resistance background of P. falciparum is only one of several determinants for in vivo treatment outcome. Whereas acquired immunity can account for the clearance of drug resistant genotypes, diminished drug metabolism may well explain treatment failure in spite of an infection with a susceptible genotype .
Regarding former drug history in PNG, the relevance of key pfdhps mutations in predicting treatment failure was expected. AQ and CQ as inefficacious partner drugs of SP in the new standard regimen were not able to curb both, the progression of pyrimethamine resistance as well as the emergence of sulphadoxine resistance. It is most likely that in this sample, clinical efficacy of the sulpha component was mainly assessed. However, according to the present results, also pfmdr1 N86Y plays an important role in predicting a negative treatment response. CQ and AQ are chemically related drugs and cross-resistance has been described in several clinical and in vitro reports. Though little is known about the genetic mechanisms conferring AQ resistance , an important role has been ascribed to the key CQR markers pfcrt K76T and pfmdr1 N86Y [39, 40]. It has been shown recently that in combination with pfcrt K76T, the pfmdr1 N86Y polymorphism was predictive for treatment failure with AQ in Nigeria  and that AQ resistance was associated with the selection of these polymorphisms in Kenya . Considering the long use of AQ as monotherapy against uncomplicated falciparum malaria in PNG and the observation that pfmdr1 N86Y is a strong predictor for treatment failure with AQ+SP, the present data support the hypothesis that pfmdr1 N86Y is probably involved in AQ resistance. Several studies have shown that both, SNPs and gene amplification of pfmdr1, can mediate resistance to 4-aminoquinlines and also other drug classes, such as amino alcohols and artemisinin derivates  However, results from different studies investigating the relationship of these genetic alterations in pfmdr1 and in vivo response were often inconsistent [6, 44]. Direct (active drug translocation) and indirect (modification of biophysical cell parameters) modes of action have been proposed for P-glycoprotein homolog 1, the gene product of pfmdr1. But how genetic alterations in pfmdr1 and epistatic interactions with other genes finally lead to a multidrug resistant phenotype remains to be resolved .
Finally, the fact that the difference in clinical outcome between the two sites was reflected in the genetic profile of the corresponding parasite populations, especially for the frequencies of pfmdr1 N86Y and pfdhps A437G, further confirmed the role of these two markers as important predictors for a negative treatment response with AQ+SP and suggests them to be the most useful resistance surveillance markers with the current standard treatment in PNG.