Polymerase Chain Reaction Adjustment in Antimalarial Trials: Molecular Malarkey?

Malaria drug trials use both clinical and microbiologic end points to establish efficacy. The recurrence of parasitemia after treatment is an indicator of treatment failure. The study by Faucher et al. [1] from Benin in this issue of the Journal demonstrates both substantial in vivo resistance to sulfadoxine‐pyrimethamine (SP) and good effectiveness of 2 new artemisinin combination therapies (ACTs): artemether‐lumefantrine (AL) and artesunate‐amodiaquine (ASAQ). Both ACTs were markedly more effective than SP. Of note, in the per‐protocol analysis, there were significantly fewer recurrent episodes of parasitemia in the AL group than in the ASAQ group, but there was no difference after polymerase chain reaction (PCR) adjustment (often called PCR correction). Are the ACT regimens equivalent, or is AL better?. The answer hinges on whether PCR correction truly corrects. Distinguishing recrudescent infections from new infections during follow‐up bedevils the interpretation of antimalarial efficacy data. To this end, PCR adjustment is used to prevent reinfections from looking like drug failures. More than 100 antimalarial trials, indexed in PubMed, have used PCR‐adjusted results. This adjustment is used because clinical trials of antimalarial drugs are typically conducted in areas with relatively high malaria prevalence. Although a high prevalence is good for recruitment, trial participants risk acquiring a new infection during the follow‐up period by being bitten by another infected mosquito. Thus, recurrences of parasitemia after treatment that are attributable to reinfection might be mistaken for recrudescences (drug failures). This could lead to spuriously low cure rates and can mistakenly make efficacious drugs look less effective. Although it is clinically desirable to prevent both recrudescence and reinfection, recrudescence alone is the primary correlate for antimalarial efficacy. For Plasmodium falciparum malaria, an early solution to this problem was simply to monitor patients for 14 days. It was thought that, because of the 7–14‐day prepatent period (time required for the parasite to mature in the liver before entering red blood cells), a reinfection would be unlikely to lead to patent malaria infection within 14 days. However, Stepniewska et al. [2] showed that many recrudescences occurred after 14 days, and that a 28‐ or 42‐day follow‐up period was needed to catch all treatment failures. Longer follow‐up periods, however, increase the likelihood of reinfection. Accordingly, to measure efficacy in trials with 28‐ or 42‐day follow‐up periods, a better means of accurately identifying recrudescence was needed. PCR correction of P. falciparum trial results was introduced in the 1990s by Snounou and Beck [3] and others as a means to better distinguish reinfections from recrudescences. Subsequently, a variety of genotyping methods and interpretations was introduced [4]. To formalize the laboratory protocols and interpretation of data, the World Health Organization and Medicines for Malaria Venture recently published a suggested standard operating procedure [5]. This method involves comparing size polymorphisms of variable surface antigens from parasites by PCR amplification before treatment and during recurrence. Amplicon length can be assessed by either gel or capillary electrophoresis. Because amplicons of different sequences or slightly different sizes may look identical, PCR amplification of 3 polymorphic genetic loci (msp1, msp2, and glurp) is recommended. If the pretreatment and recurrent infections share ⩾1 band in all 3 loci, the recurrence is diagnosed as a recrudescence (figure 1A). If no bands are shared at any locus, the recurrence is considered to be a reinfection (figure 1B). Despite this standardization, multiple factors compromise the ability of PCR correction to evaluate recurrent parasitemia robustly and identify recrudescences, resulting in false underestimation and overestimation of antimalarial clinical efficacy. Representative banding patterns for polymerase chain reaction (PCR) correction. Panels show typical banding patterns from pretreatment (Pre) and posttreatment (Post) samples by the World Health Organization and Medicines for Malaria Venture guidelines for PCR correction of antimalarial clinical trials. A, Pattern that would be interpreted as a recrudescence because of a shared band before and after treatment. B, Pattern for a reinfection with no shared alleles. A major drawback of PCR correction is its inability to capture fully the polyclonality of Plasmodia infection. Recrudescences can be misclassified as reinfections, primarily because of in‐host diversity. Single individuals with malaria may harbor >10 genetically distinct variants, and some variants (“minority variants”) may be present at very low levels. PCR amplification of mixtures of alleles cannot reliably demonstrate alleles present in <20% of the population; therefore, some variants from an infected patient are cryptic and are not represented in the final amplification products [6, 7]. A cryptic drug‐resistant minority variant present at enrollment could cause recrudescence but would be visible only in the recurrent sample. Thus, recrudescences could be falsely classified as reinfections. The intensity of this effect will vary with transmission intensity. In areas where transmission intensity is high and parasite genetic diversity is great (both in the patient and in the mosquito population), minority variants are likely to be more common than in less genetically diverse, low‐transmission areas. As a result, PCR correction can lead to overestimates of drug efficacy. Reinfections can also be misdiagnosed as recrudescences. In any geographic region, a few variants may be much more common than others. If a patient carried a common variant at enrollment, there is a very good chance that he or she was reinfected with the same variant during the follow‐up period. Indeed, 2 studies from Africa estimated that ∼40% of PCR‐adjusted recrudescences were actually reinfections [8, 9]. In areas of low transmission, where the average complexity of infection is low and the genetic diversity of parasites in mosquitoes is limited, the likelihood of this type of misclassification is even higher. Reinfections may also be misdiagnosed as recrudescences because of the limited ability of current techniques to detect unique alleles. PCR adjustment protocols will detect size polymorphisms but not sequence polymorphisms. In addition, they are unable to differentiate small size polymorphisms. Two genetically distinct variants whose PCR products are close in size, 1 variant present at enrollment and 1 at recurrence, may be misidentified as the same variant, leading to the misclassification of a reinfection as a recrudescence. Because of both of these biases, the efficacy of a drug regimen may be underestimated with PCR correction. In the end, is PCR correction a good idea?. Undoubtedly, we need the tools to accurately diagnose recrudescent parasitemia as a means to monitor drug efficacy and resistance. Although it may be possible to compensate statistically for transmission rates, to merge or compare results at different sites [8–10], the technological inability of current PCR correction strategies to detect minority variants and distinguish alleles that are close in size remains problematic. Newer technologies, such as heteroduplex tracking assays or ultradeep sequencing by massively parallel pyrosequencing, could be used for diagnosis of recrudescence [9, 11, 12]. These technologies may yield a more sophisticated and reliable test than the crude PCR correction. Until that time, we cannot conclusively interpret the puzzling results of studies such as that of Faucher et al. [1]. Therefore, policy makers need to look at both the raw and PCR‐corrected data before making recommendations.